CN101656882B - Moving image coding method, and moving image decoding method - Google Patents

Abstract

A moving image coding apparatus (10) for performing inter-picture predictive coding on pictures constituting a moving image is provided with a coding unit (103) for performing predictive error coding on image data; a decoding unit (105) for performing predictive error decoding on an output from the coding unit (103); a reference picture memory (117) for holding output data from the decoding unit (105); and a motion vector detection unit (108) for detecting forward and backward motion vectors on the basis of decoded image data stored in the memory (117). When coding a B picture, a picture that is timewise closest to the target picture is used as a candidate picture for forward reference, and an I or P picture that is timewise closest to the target picture is used as a candidate picture for backward reference. The moving image coding apparatus (10) so constructed can improve coding efficiency of a B picture to be subjected to bidirectional predictive coding.

Description

Dynamic image encoding method and dynamic image decoding method

The application is that application number is 02815766.4, the applying date is on November 6th, 2002, denomination of invention is divided an application for " dynamic image encoding method and dynamic image decoding method ".

Technical field

The present invention relates to moving image encoding (Video coding) method and moving picture decoding (video decode) method, relate in particular to consisting of each picture of moving image, the method for encoding or decoding with reference to other pictures of this moving image.

Background technology

Generally, when the picture that consists of moving image was encoded, each picture was divided into a plurality of, and utilizes the direction in space of moving image and the redundancy of time direction, to every compressed encoding that carries out the image information that each picture has (below, also referred to as coding).As the coding of the redundancy of utilizing direction in space, there is the relevant interior coding of picture of pixel value that utilizes in the picture.As the coding of the redundancy of utilizing time orientation, there is the relevant inter-picture prediction coding of pixel value that utilizes between picture.The inter-picture prediction coding is, to (to be encoded) object picture, be positioned in time the picture (forward direction picture or forward direction are with reference to picture) in the place ahead with reference to relative this object picture, perhaps this object picture is positioned at the processing that the picture (backward picture or backward with reference to picture) at rear is encoded in time relatively.

The forward direction picture is picture more Zao than object picture on DISPLAY ORDER, on the time shaft (below, be called " displaying time axle ") of the displaying time of each picture of expression, is positioned at the front side of object picture.Backward picture is picture slower than object picture on DISPLAY ORDER, on the displaying time axle, is positioned at the rear side of object picture.In addition, in the following description, be called with reference to picture at the picture of time institute's reference that the object picture is encoded.

In the inter-picture prediction coding, specifically be exactly by the motion vector of detected object picture with respect to the reference picture, based on the motion compensation process of this motion vector, to obtain the prediction data of the view data of this object picture.Then, the redundancy on the direction in space of the picture of the differential data of the view data by removing this prediction data and object picture is carried out compressed encoding for the data volume of object picture.

In addition, process as the decoding of the picture of encoding, the interior decoding of picture of coding in the corresponding picture arranged, and decode between the picture of encoding between corresponding picture.Between picture in the decoding, with reference to picture between the identical picture of picture of coding time institute's reference.Namely, with reference to picture Xra and the coded picture Xtg of Xrb, decode with reference to above-mentioned picture Xra and Xrb.

Figure 43 (a)-43 (c) is the figure that expression consists of a plurality of pictures of moving image.

In Figure 43 (a), the part of a plurality of pictures that consist of a moving image Mpt is shown, namely, picture F (k)～F (k+2n-1) [k, n: integer].In each picture F (k)～F (k+2n-1), be set with displaying time t (k)～t (k+2n-1).Shown in Figure 43 (a), this each picture, on the displaying time axle X of each displaying time Tdis of expression, it is arranged sequentially to press displaying time, these pictures, (n) is divided into group by each predetermined number.These picture groups are called as GOP (Group of Pictures), become for the coded data of moving image and carry out random-access least unit.In the following description, the picture group is sometimes also noted by abridging and is GOP.

For example, (i) individual picture group Gp (i) is made of picture F (k)～F (k+n-1).(i+1) individual picture group Gp (i+1) is made of picture F (n+k)～F (k+2n-1).

Each picture is divided into the picture bar (slice) that consists of a plurality of macro blocks.For example, macro block is all to be 16 rectangular area in the vertical direction with the pixel count of horizontal direction.In addition, picture F (k+1) shown in Figure 43 (b), is divided into a plurality of picture bar SL1～SLm[m: natural number].Picture bar SL2 is shown in Figure 43 (c), by a plurality of macro block MB1～MBr[r: natural number] consist of.

Figure 44 is the figure for the coded data of account for motion image, and expression is to the encode structure of resulting stream of each picture that consists of moving image.

Stream Smp is the coded data of a corresponding image sequence (for example moving image).This stream Smp is by zone (common information zone) Cstr of the bit string that is arranged with the common information that is equivalent to title etc., and zone (GOP zone) Dgop that is arranged with the bit string of corresponding each GOP consists of.In the Cstr of common information zone, include synchrodata Sstr and title Hstr corresponding to stream.In the Dgop of GOP zone, include the bit string Bg (1) of corresponding picture group (GOP) Gp (1)～Gp (i-1), Gp (i), Gp (i+1)～Gp (I) [i, I: integer]～Bg (i-1), Bg (i), Bg (i+1)～Bg (I).

Corresponding to each bit string of each GOP, by zone (common information zone) Cgop of the bit string that is arranged with the common information that is equivalent to title etc., and zone (picture area) Dpct that is arranged with the bit string of corresponding each picture consists of.In the Cgop of common information zone, include synchrodata Sgop and title Hgop corresponding to GOP.In the picture area Dpct of the bit string Bg of corresponding picture group G (i) (i), include corresponding picture F (k '), F (k '+1), F (k '+2), F (k '+3) ..., F (k '+s) [k ', s: integer] bit string Bf (k '), Bf (k '+1), Bf (k '+2), Bf (k '+3) ..., Bf k '+s).Picture F (k '), F (k '+1), F (k '+2), F (k '+3) ..., F (k '+s) change by coded sequence and arranged the picture F (k) that arranges by above-mentioned DISPLAY ORDER～F (k+n-1).

Each bit string of corresponding each picture by zone (common information zone) Cpct of the bit string that is arranged with the common information that is equivalent to title etc., and is arranged with corresponding each zone as the bit string of bar (as the bar zone) Dslc and consists of.In the Cpct of common information zone, include synchrodata Spct and title Hpct for picture.For example, in the situation of picture F (k+1) of (DISPLAY ORDER arrangement) arranged sequentially by displaying time at the picture F (k '+1) of (coded sequence arrangement) arranged sequentially by the scramble time, in the picture bar zone Dslc of the bit string Bf (k '+1) of corresponding picture F (k '+1), comprise corresponding above-mentioned each bit string Bs1 as bar SL1～SLm～Bsm.

Corresponding each each bit string as bar, by zone (common information zone) Cslc of the bit string that is arranged with the common information that is equivalent to title etc., and zone (macro block zone) Dmb that is arranged with the bit string of corresponding each macro block consists of.In the Cslc of common information zone, include synchrodata Sslc and title Hslc for the picture bar.For example, be in the situation of the picture F (k+1) that arranges of DISPLAY ORDER at the picture F (k '+1) that coded sequence is arranged, in the macro block zone Dmb of correspondence as the bit string Bs2 of bar SL2, comprise the bit string Bm1 of corresponding above-mentioned each macro block MB1～MBr～Bmr.

As mentioned above, the coded data of a corresponding moving image (namely image sequence), have by correspondence as the fluid layer of the stream Smp of this coded data, GOP layer that correspondence consists of above-mentioned stream, the picture frame layer and being equivalent to that is equivalent to consist of the picture of above-mentioned each GOP consists of the hierarchical structure that the picture bar layer of the picture bar of above-mentioned each picture forms.

; MPEG (Moving Picture Experts Group)-1, MPEG-2, MPEG-4, ITU-T suggestion H.263, in the moving image encoding mode that H.26L waits, the picture that the picture of implementing coding in the picture is called as the I picture, implement the inter-picture prediction coding is called as P picture or B picture.

Below, the definition of I picture, P picture, B picture is described.

The I picture is not with reference to the coded picture of other picture.P picture or B picture are with reference to the coded picture of other picture.Strictly, the P picture is when each picture is encoded, and can select any one picture that the I mode is encoded and the P mode is encoded.The B picture is when each picture is encoded, and can select any one picture that the I mode is encoded, the P mode is encoded and the B mode is encoded.

The I mode is encoded, and is the object piece of object picture not to be carried out the processing of encoding in the picture with reference to other picture.The P mode encode be with reference to one coded picture the object piece of object picture is carried out the processing of inter-picture prediction coding.The B mode encode be with reference to two coded picture the object piece of object picture is carried out the processing of inter-picture prediction coding.

Encoding or the encode picture of time institute's reference of B mode in above-mentioned P mode, is the offscreen I picture of object or P picture, also can be the forward direction picture that is positioned at the place ahead of object picture, and be positioned at the backward picture at the rear of object picture.

But, in the B mode is encoded, carry out in two pictures and combination of reference three kinds of situations being arranged.Namely, the situation with reference to two forward direction pictures in encoding, the B mode is arranged, the situation of two backward pictures, and the situation of a forward direction picture of reference and a backward picture.

Figure 45 is the figure of the moving image encoding mode of explanation above-mentioned MPEG etc.In Figure 45, the object picture is shown, and the corresponding relation with reference to picture (at the picture of time institute's reference that the object picture is encoded).

Consist of each picture F (k) of moving image～F (k+7) ..., the coding of F (k+17)～F (k+21), the picture with reference to other carries out like that shown in arrow Z.Specifically, being in the picture of the end side of an arrow Z, is to encode by inter-picture prediction with reference to the picture that is in the top side of this arrow.In Figure 45, above-mentioned picture F (k)～F (k+7) ..., the picture F (k) shown in F (k+17)～F (k+21) and Figure 43 (a)～F (k+4) ..., F (k+n-2)～F (k+n+4) ..., F (k+2n-2), F (k+2n-1) be identical.These pictures are pressed DISPLAY ORDER at displaying time axle X and are arranged.Picture F (k)～F (k+7) ..., the displaying time of F (k+17)～F (k+21), be time t (k)～t (k+7) ..., t (k+17)～t (k+21).The picture type of picture F (k)～F (k+7) is, I, B, B, P, B, B, P, B, and the picture type of picture F (k+17)～F (k+21) is, B, P, B, B, P.

For example, from the B mode of the B picture F (k+1) of second of beginning shown in Figure 45 is encoded, with reference to the I picture F (k) of beginning and the 4th P picture F (k+3).In addition, from the P mode of the P picture F (k+3) of the 4th of beginning shown in Figure 45 is encoded, with reference to the I picture F (k) of beginning.

Although shown in Figure 45 in the P of P picture mode is encoded with reference to the situation of forward direction picture, also can be with reference to rear to picture in the P of P picture mode is encoded.In addition, although shown in Figure 45 in the B of B picture mode is encoded with reference to forward direction picture and rear situation to picture, also can be with reference to two forward direction pictures in the B of B picture mode is encoded, perhaps with reference to two backward pictures.

In addition, in MPEG-4 and the moving image encoding mode that H.26L waits, when the B picture is encoded, can select to be called the coded system of direct mode.

Figure 46 (a) and Figure 46 (b) are the figure that encodes for the inter-picture prediction that explanation is carried out with this direct mode.Shown in Figure 46 (a) in direct mode employed motion vector.

In Figure 46 (a), picture P1, B2, B3, P4 are equivalent to picture F (k+3) shown in Figure 45～F (k+6) [k=-2], and time t (1), t (2), t (3), t (4) (t (1)＜t (2)＜t (3)＜t (4)) are the displaying times of picture P1, B2, B3, P4.In addition, X is the displaying time axle of expression displaying time Tdis among the figure.

Below, describe particularly with regard to the situation of in direct mode the piece BL3 of picture B3 being encoded.

In the case, becoming coding, to process the object picture of object be picture B3, and the object piece that becomes the object that coding processes is piece BL3.

In the predictive coding of the piece BL3 of picture B3 is processed, utilize the rear that is positioned at picture B3, the motion vector MV4 of the piece BL4 of the picture P4 that encodes recently.This piece BL4 equates for the relative position of picture B3 with above-mentioned BL3 for the relative position of picture P4.Namely, shown in Figure 46 (b), with respect to the coordinate (x4, y4) of the datum mark Ob4 of the piece BL4 of the datum mark O4 of picture P4, equate with coordinate (x3, y3) with respect to the datum mark Ob3 of the piece BL3 of the datum mark O3 of picture P3.In addition, the motion vector MV4 of piece BL4 is employed motion vector in the predictive coding of piece BL4.The motion vector MV4 of this piece BL4 is that the motion detection of the piece BL4 by reference forward direction picture P1 obtains, the regional R4f of expression forward direction picture P1 corresponding blocks BL4.

Then, for the piece BL3 of above-mentioned picture B3, use motion vector MV3f and the MV3b parallel with motion vector MV4, implement the bi-directional predictive coding with reference to forward direction picture P1 and backward picture P4.Motion vector MV3f is illustrated in encode regional R3f forward direction picture P1, corresponding blocks BL3 of time institute's reference to piece BL3.Motion vector MV3b is illustrated in the encode regional R3b backward picture P4, corresponding blocks BL3 of time institute's reference to piece BL3.

, in ITU-T suggestion (Annex U H.263++), record and narrate to some extent about the framework in the situation that a plurality of pictures are used as the candidate of reference picture.In this records and narrates, be used for preserving become with reference to the view data of the picture (candidate screen) of the candidate of picture with reference to picture memory, be divided into short period picture memory and long period picture memory.The short period picture memory is to preserve the in time data storage zone of near candidate screen (adjacent candidate screen) of relative object picture.The long period picture memory is to preserve the in time data storage zone of candidate screen (distant place candidate screen) far away of relative object picture.Specifically, above-mentioned distant place candidate screen, the distance with the object picture leaves on the quantity of the candidate screen from the object picture to this distant place candidate screen, surpasses the quantity that can be stored in the candidate screen in the short period picture memory.

In addition, in ITU-T suggestion (AnnexU H.263++), the record that utilizes method relevant for short period picture memory and long period picture memory, and and then, about picture being specified the also to some extent record of method with reference to the index (following also referred to as cross index) of picture.

At first, just the designation method for the cross index of picture describes simply.

Figure 47 (a) and Figure 47 (b) specify the figure of the method for cross index for explanation to a plurality of pictures that consist of moving image.Figure 47 (a) is illustrated in the be encoded candidate (candidate screen) of picture of time institute's reference of picture P16.Figure 47 (b) is illustrated in the be encoded candidate (candidate screen) of picture of time institute's reference of picture B15.

Picture P4, B2, B3, P7, B5, B6, P10, B8, B9, P13, B11, B12, P16, B14, B15, P19, B17, P18 shown in Figure 47 (a) changes by coded sequence and to have arranged picture F (k+1) shown in Figure 45～F (k+17) and (k=1) obtain.The arrangement of a plurality of pictures shown in Figure 47 (a) is arrangements of the picture on time shaft (scramble time axle) Y of time (scramble time) Tenc that expression is encoded to each picture.

Below, shown in Figure 47 (a), the situation that the piece enforcement P mode of P picture P16 is encoded describes.

In the case, in 4 forward direction P pictures (picture P4, P7, P10, P13), be suitable for the picture that carries out reference in when coding with reference to one.Namely, above-mentioned forward direction P picture P4, P7, P10, P13 are the candidate screen that can be designated as when the P of picture P16 mode is encoded with reference to picture.Give respectively cross index to these candidate screen P4, P7, P10, P13.

For in the giving of the cross index of such candidate screen, on above-mentioned scramble time axle Y, apart from the nearer candidate screen of picture P16 that becomes coded object, the cross index that the value of being endowed is less.Specifically, be exactly shown in Figure 47 (a), on to picture P13, P10, P7, P4, give respectively cross index [0], [1], [2], [3].In addition, the information of the cross index that expression is given each candidate screen as the parameter of motion compensation, is recorded and narrated in the bit string of the object piece of corresponding picture P16.

Then, shown in Figure 47 (b), the situation that the piece enforcement B mode of B picture B15 is encoded describes.

In the case, in 4 forward direction pictures (picture P4, P7, P10, P13) and 1 backward picture (picture P16), with reference to two pictures that carry out reference when being suitable for encoding.Namely, forward direction picture P4, P7, P10, P13 and backward picture P16 are the candidate screen that can be designated as when the B of picture B15 mode is encoded with reference to picture.Behind 4 forward direction pictures and 1, become in the situation of candidate screen to picture, give cross index to forward direction picture P4, P7, P10, P13, and to backward picture P16 give expression it by the code of the candidate screen of backward reference [b].

Carrying out in the giving of cross index for such candidate screen, to the forward direction picture as candidate screen, on above-mentioned scramble time axle Y, the picture B15 of coded object is nearer apart from becoming, just the less cross index of the value of giving.Specifically, be exactly shown in Figure 47 (b), on picture P13, P10, P7, P4, give cross index [0], [1], [2], [3].In addition, the information of the cross index that expression is given each candidate screen as the parameter of motion compensation, is recorded and narrated in the bit string of the object piece of corresponding picture B15.

Then, will advise at ITU-T recording and narrating in (AnnexU H.263++), the adding method of cross index, with short period picture memory and long period picture memory utilize method to associate to describe.

In the short period picture memory, storage can be used as the candidate screen of carrying out appointment with reference to picture for the object picture in order, and to the candidate screen of storing by give cross index (in other words, according to decoding order, or according to the bit string order) to the storage order of memory.In addition, when the B picture was decoded, the picture of storing in memory was recently treated with reference to picture as backward, and to other picture, carried out giving of cross index with the storage order to memory.

Below, just can be used as for the candidate with reference to picture of object picture and utilize the situation of 4 forward direction pictures to describe particularly.

Figure 48 (a) and Figure 48 (b) are the figure of the part of expression a plurality of pictures of consisting of moving image, wherein represent the situation of picture having been carried out arranging (Figure 48 (a)) and by coded sequence picture having been carried out arranging (Figure 48 (b)) by DISPLAY ORDER.In addition, picture P1, B2, B3, P4, B5, B6, P7, B8, B9, P10, B11, B12, P13, B14, B15, P16, B17, B18, P19 shown in Figure 48 (a) is equivalent to picture F (k+3) shown in Figure 45～F (k+21) [k=-2].

Figure 49 is for the figure with reference to the management example of picture memory of explanation for the picture of such arrangement.

In Figure 49, when being encoded, the object picture is kept at reference to the coded picture in the picture memory, with corresponding this logical storage numbering of the storage area of coded picture of preserving, and to this coded picture cross index of giving associate and represent.

In Figure 49, picture P16, B14, B15 are the object pictures.Logical storage numbering (0)～(4), expression is with reference to the position in logic (storage area) in the picture memory.Be encoded time of (perhaps decoding) of the value of the logical storage numbering of corresponding each storage area, the processing picture of storing in storage area is slower, just becomes less value.

Below, the management of reference picture memory is described particularly.

When the coding (decoding) of picture P16, in the storage area reference picture memory, (0)～(3) expression of logical storage numbering, store respectively picture P13, P10, P7, P4.Give respectively cross index [0], [1], [2], [3] to this picture P13, P10, P7, P4.

When the coding (decoding) of picture B14 and B15, in the storage area reference picture memory, (0)～(4) expression of logical storage numbering, picture P16, P13, P10, P7, P4 are arranged.At this moment, give expression as by the code of the candidate screen of backward reference [b] to picture P16, and to remaining by the candidate screen P13 of forward direction reference, P10, P7, P4, give respectively cross index [0], [1], [2], [3].

The information of the reference marker that expression is given above-mentioned each candidate screen, it is the parameter of motion compensation, when the piece of object picture is encoded, will which picture in above-mentioned a plurality of candidate screen have carried out the information of using as the reference picture as expression, record is in should the bit string of piece.At this moment, in this cross index, its value is less just gives shorter code.

; in above-mentioned existing coding method, because when the piece for the B picture carried out predictive coding, I picture or P picture were designated as and are the reference picture; so object picture and can be elongated with reference to the distance of picture on the displaying time axle (below, be also referred to as time gap).

For example, when in the predictive coding for the piece of B picture B15 shown in Figure 48 (b), forward direction picture P13 and backward picture P16 are designated as in the situation into the reference picture, time gap Ltd between B picture B15 (object picture) and the forward direction picture P13 (with reference to picture) (=t (15)-t (13)), shown in Figure 50 (a), become 2 picture interval (2Pitv).

In addition, when in predictive coding shown in Figure 48 (b), for the piece of B picture B15, forward direction picture P13 and P10 are designated as in the situation with reference to picture, time gap Ltd between B picture B15 (object picture) and the forward direction picture P10 (with reference to picture) (=t (15)-t (10)), shown in Figure 50 (b), become 5 picture interval (5Pitv).

Especially when between I picture and P picture, the number of the B picture that perhaps inserts between two of adjacency P pictures becomes in the many situation, object picture and then become greatly with reference to the time gap Ltd between the picture, and this just becomes significantly low reason of code efficiency.

In addition, in common coding method, have in can encoding with reference to a plurality of rear B modes to picture, to distance object picture near picture nearby in time, give than the distant place picture far away in time of this object picture certainly, also want the situation of the cross index of large value.

In the case, in the motion detection for the piece of object picture, because closer candidate screen situation about being referenced is more in time from the object picture on the probability, in other words, since apart from the object picture in time closer candidate screen be designated as with reference to the situation of picture more, so cause the deteriorated of code efficiency.

Below, just in the B mode for the piece of B picture B15 shown in Figure 51 (a) is encoded, describe particularly in the situation that two backward picture P16 and P19 are referenced.

In the case, shown in Figure 51 (a), press picture B2, B3, P4, B5, B6, P7, B8, B9, P10, B11, B12, P13, B14, B15, P16, B17, B18, P19 that DISPLAY ORDER is arranged, changed row to become coded sequence P7, B2, B3, P10, B5, B6, P13, B8, B9, P16, B11, B12, P19, B14, the B15 shown in Figure 51 (b).

In addition, in the case, in 3 forward direction pictures (picture P7, P10, P13) and 2 backward pictures (picture P16, P19), with reference to two pictures that are suitable for encoding and processing.Specifically, forward direction picture P7, P10, P13, and backward picture P16, P19 are the candidate screen that can be designated as when the piece of picture B15 is encoded with reference to picture.Behind 3 forward direction pictures and 2, to picture, as mentioned above, become in the situation of candidate screen, give cross index to forward direction picture P7, P10, P13 and backward picture P16, P19.

For in the giving of the cross index of such candidate screen, on above-mentioned scramble time axle Y, apart from the nearer candidate screen of picture B15 that becomes coded object, the cross index that the value of being endowed is less.Specifically, be exactly shown in Figure 51 (b), to picture P19, P16, P13, P10, P7, give respectively cross index [0], [1], [2], [3], [4].

; in the case, on displaying time axle X, the value [1] of the cross index that is endowed apart from the near P picture P16 of object picture (picture B15); just large as the value [0] of the cross index of the B picture B15 of object picture P picture P19 far away than oneself, just cause the deteriorated of code efficiency.

The present invention finishes in order to solve problem points as described above, purpose provides the dynamic image encoding method that can avoid by the reduction of object picture and and the code efficiency that cause long with reference to the time gap between the picture, and corresponding to the dynamic image decoding method of the dynamic image encoding method of the reduction that can avoid like this code efficiency.

In addition, the purpose of this invention is to provide the candidate screen that can be referenced in predictive coding is processed, can not cause the deteriorated of code efficiency and give the dynamic image encoding method of cross index, and corresponding to the dynamic image decoding method of the deteriorated dynamic image encoding method that can avoid like this code efficiency.

Summary of the invention

According to dynamic image encoding method of the present invention, that a kind of a plurality of pictures consisting of moving image are divided into respectively a plurality of, by the dynamic image encoding method that each piece is encoded to each picture, the method comprises the coding step that the piece of the object picture that becomes coded object is carried out predictive coding with reference to coded picture; And above-mentioned coding step is when coded picture has carried out the B picture of predictive coding of piece of this object picture with reference to two at the object picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, B picture of having encoded of reference carries out predictive coding to the piece of this object picture, and be when coded picture has carried out the P picture of predictive coding of its piece with reference to one at the object picture, with reference to any B picture of having encoded each piece of this object picture do not carried out predictive coding.

According to the present invention, in above-mentioned dynamic image encoding method, in each of a plurality of pictures that consist of above-mentioned moving image, respectively in above-mentioned coding step as the I picture that does not carry out the coding of its piece with reference to coded picture, with reference to the coded picture P picture that carries out the predictive coding of its piece, and with reference to two the coded picture a certain picture of B picture that carries out the predictive coding of its piece be carried out coding; And, above-mentioned coding step, when the object picture is the B picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding, and the candidate screen number with reference to picture of reference when this object picture is encoded as the B picture is equal to or less than the candidate number with reference to picture of reference when this object picture encoded as the P picture.

According to the present invention, in above-mentioned dynamic image encoding method, in each of a plurality of pictures that consist of above-mentioned moving image, respectively in above-mentioned coding step as the I picture that does not carry out the coding of its piece with reference to coded picture, with reference to the coded picture P picture that carries out the predictive coding of its piece, and with reference to two the coded picture a certain picture of B picture that carries out the predictive coding of its piece be carried out coding; And, above-mentioned coding step, when the object picture is the B picture, the B picture that the piece of this object picture is carried out the reference of predictive coding time institute, just this object picture and relatively this object picture at the B picture between the immediate I in position or the P picture on the displaying time axle.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding, and the picture position information that represents the position of B picture on the displaying time axle of having encoded of this reference is included in by in the resulting bit string that the picture that consists of above-mentioned moving image is encoded.

According to the present invention, in above-mentioned dynamic image encoding method, above-mentioned picture position information, by from above-mentioned object picture to nearer in the distance B picture of having encoded of time institute's reference that it is encoded, on the displaying time axle, code length just shorter code is represented.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, the piece of this object picture is carried out predictive coding and in by the resulting bit string of the coding of the picture that consists of above-mentioned moving image with reference at least one B picture of having encoded, comprise the information that is illustrated in the B picture of having encoded of reference when encoding as the B picture of above-mentioned object picture as heading message.

According to dynamic image encoding method of the present invention, that each of a kind of a plurality of pictures consisting of moving image is divided into respectively a plurality of, the dynamic image encoding method of each picture being encoded by each piece, the method comprise at least with reference to one coded picture carry out with reference to and implement the P picture of the predictive coding of its piece, with with reference to two coded picture B pictures of carrying out the predictive coding of its piece, the coding step that the object picture that becomes coded object is encoded; And, in above-mentioned coding step, be arranged in use this object picture reference pictures of having encoded nearby, be positioned at and this object piece direct mode of the motion vector of the reference block of identical position spatially, when encoding as the object piece of the B picture of object picture, with reference to according to the determined coded picture of certain rule.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to being positioned near before the above-mentioned object picture and the DISPLAY ORDER first coded picture more Zao than this object picture.

According to the present invention, in above-mentioned dynamic image encoding method, comprise the reference pictures of having encoded of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned coding step, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and this reference block be encoded the second picture of time institute's reference on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the first picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the second picture on the displaying time axle apart from TRB, obtain the forward motion vector (MVR * TRF/TRD) of above-mentioned object piece, and backward motion vector ((TRB-TRD) * MVR/TRD), and use this forward motion vector and this backward motion vector to carry out bi-directional predicted.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to the coding forward direction picture the most nearby, that DISPLAY ORDER is more Zao than this object picture that is positioned at above-mentioned object picture, and with reference to being positioned at the backward picture of coding, that DISPLAY ORDER is slower than this object picture of above-mentioned object picture the most nearby, the motion vector that carries out above-mentioned object piece be 0 bi-directional predicted.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, the image information of economizing the whereabouts prediction error information and be 0 object piece is inserted in, bit string corresponding to above-mentioned moving image.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, be 0 o'clock in the prediction error information of above-mentioned object piece, the image information of saving this object piece is corresponding to the insertion in the bit string of above-mentioned moving image.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, to becoming the candidate screen with reference to picture of when the object picture is encoded, carrying out reference, give with reference to picture index, when in direct mode the object piece being encoded, with reference to being endowed specific candidate screen with reference to picture index.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, near the picture before the object picture, only when the object picture coding in the situation as the candidate screen of reference picture, when this object picture coding, carry out in the candidate screen of reference, to except near the picture that is positioned at the place ahead of this object picture the picture before this object picture, give above-mentioned specific with reference to picture index.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture codec, carry out in the candidate screen of reference, more Zao than this object picture to DISPLAY ORDER, except the candidate screen near the most close offscreen this object picture before this object picture, give above-mentioned specific with reference to picture index.

According to the present invention, in above-mentioned dynamic image encoding method, comprise the reference pictures of having encoded of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned coding step, when in direct mode the object piece being encoded, with reference to the first forward direction picture time institute's reference, that DISPLAY ORDER is more Zao than this object picture of encoding at the said reference piece.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to being positioned near DISPLAY ORDER than the second forward direction picture before the Zao object picture of this object picture, and, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the second forward direction picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), the backward motion vector ((TRB-TRD) * MVR/TRD) of the motion vector of above-mentioned object piece.

According to the present invention, in above-mentioned dynamic image encoding method, in above-mentioned coding step, in direct mode to above-mentioned object piece coding the time, at the DISPLAY ORDER forward direction picture for the treatment of reference more Zao than above-mentioned object picture, be not present in and keep with reference in the situation in the memory of picture, the picture of, the most close this object picture more Zao than above-mentioned object picture with reference to DISPLAY ORDER.

According to dynamic image encoding method of the present invention, be that a kind of each to a plurality of pictures of consisting of moving image is encoded and generated dynamic image encoding method to bit string that should each picture, the method comprises the coding step of object picture to be encoded being encoded with reference to the picture of having encoded; And, above-mentioned coding step comprises: to as above-mentioned to encode candidate a plurality of with reference to candidate screen of picture of time institute's reference of object picture, the picture that the DISPLAY ORDER of its DISPLAY ORDER and this coded object picture is nearer, step given in the index with reference to picture index that the value of giving is less; And will above-mentioned to the object picture encode that the picture of time institute's reference obtains to give with reference to picture index, be additional to the indexing step of above-mentioned bit string.

According to dynamic image encoding method of the present invention, be that a kind of each to a plurality of pictures of consisting of moving image is encoded and generated dynamic image encoding method to bit string that should each picture, the method comprises that the picture of having processed with reference to having encoded processes the coding step that the object picture of object is encoded to becoming coding; And, in above-mentioned coding step, in above-mentioned bit string, record and narrate the sign whether above-mentioned object picture of expression uses as the candidate of reference picture when processing for the coding of other later pictures of this object picture.

According to dynamic image decoding method of the present invention, a kind of respectively to each of a plurality of pictures of consisting of moving image, decode by each piece as the processing unit of this picture, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, comprise with reference to decoded picture the piece of object picture to be decoded is carried out the decoding step of prediction decoding; And, in above-mentioned decoding step, be when decoded picture has been carried out the B picture of prediction decoding of its piece with reference to two at the object picture, with reference at least one decoded B picture the piece of this object picture is carried out prediction decoding.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, decoded B picture of reference carries out prediction decoding to the piece of this object picture, and be when decoded picture has been carried out the P picture of prediction decoding of its piece with reference to one at the object picture, with reference to any decoded B picture each piece of this object picture do not carried out prediction decoding.

According to the present invention, in above-mentioned dynamic image decoding method, in each of a plurality of pictures that consist of above-mentioned moving image, in above-mentioned decoding step as the I picture that does not carry out the decoding of its piece with reference to decoded picture, with reference to one decoded picture carry out the P picture of the prediction decoding of its piece, and with reference to two the decoded picture a certain picture of B picture that carries out the prediction decoding of its piece be carried out decoding; And, in above-mentioned decoding step, when the object picture is the B picture, with reference at least one decoded B picture the piece of this object picture is carried out prediction decoding, and the candidate number with reference to picture of reference when this object picture is decoded as the B picture is equal to or less than the candidate number with reference to picture of reference when this object picture decoded as the P picture.

According to the present invention, in above-mentioned dynamic image decoding method, in each of a plurality of pictures that consist of above-mentioned moving image, in above-mentioned decoding step as the I picture that does not carry out the decoding of its piece with reference to decoded picture, with reference to one decoded picture carry out the P picture of the prediction decoding of its piece, and with reference to two the decoded picture a certain picture of B picture that carries out the prediction decoding of its piece be carried out decoding; And, in above-mentioned decoding step, when the object picture is the B picture, to the B picture that the piece of this object picture carries out the reference of prediction decoding time institute, just insert in this object picture and relatively this object picture at the B picture between the immediate I in position or the P picture on the displaying time axle.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, based on what in above-mentioned bit string, comprise, the picture position information that represents the position of above-mentioned decoded B picture on the displaying time axle carries out the piece of this object picture being carried out the processing of prediction decoding with reference to decoded B picture.

According to the present invention, in above-mentioned dynamic image decoding method, above-mentioned picture position information, by nearer to the distance decoded B picture, on the displaying time axle of time institute's reference of decoding at it from above-mentioned object picture, code length is represented with regard to shorter code.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, based in corresponding to the bit string as the picture of the composition that consists of above-mentioned moving image, comprise, be illustrated in when the object B picture encoded with reference to the heading message of this situation of B picture of having encoded, carry out the piece of this object picture being carried out the processing of prediction decoding with reference at least one decoded B picture.

According to dynamic image decoding method of the present invention, a kind of respectively to each of a plurality of pictures of consisting of moving image, decode for each piece as the processing unit of each picture, thereby will be for conversion into corresponding to the bit string of this each picture the dynamic image decoding method of view data, the method comprise with reference at least to one decoded picture carry out with reference to and implement the P picture of the prediction decoding of its piece, with with reference to two decoded picture carry out the B picture of the prediction decoding of its piece, object picture to be decoded is carried out the decoding step of prediction decoding; And, in above-mentioned decoding step, be arranged in use this object picture decoded reference pictures nearby, be positioned at and this object piece direct mode of the motion vector of the reference block of identical position spatially, when decoding as the object piece of the B picture of object picture, with reference to according to the determined decoded picture of certain rule.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to being positioned near before the above-mentioned object picture and the DISPLAY ORDER first coded picture more Zao than this object picture.

According to the present invention, in above-mentioned dynamic image decoding method, comprise the decoded reference pictures of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned decoding step, based on the big or small MVR of the motion vector of said reference piece, above-mentioned backward reference P picture and the second picture of this reference block institute's reference of decoded time on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the first picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the second picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), and backward motion vector ((TRB-TRD) * MVR/TRD) of above-mentioned object piece.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to the decoding forward direction picture the most nearby, that DISPLAY ORDER is more Zao than this object picture that is positioned at above-mentioned object picture, and with reference to being positioned at the backward picture of decoding, that DISPLAY ORDER is slower than this object picture of above-mentioned object picture the most nearby, the motion vector that carries out above-mentioned object piece be 0 bi-directional predicted.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, use the motion vector of said reference piece, to be not included in the above-mentioned bit string, prediction error information is that the image information of 0 object piece is restored.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, use the motion vector of said reference piece, to be not included in the above-mentioned bit string, prediction error information is that the image information of 0 object piece is restored.

According to the present invention, in above-mentioned dynamic image decoding method, to carrying out the candidate screen of the picture of reference when the object picture codec, give with reference to picture index; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed specific picture with reference to picture index.

According to the present invention, in above-mentioned dynamic image decoding method, near the picture before the object picture, when the object picture is decoded, only be used as in the situation with reference to candidate screen, when this object picture codec, carry out in the candidate screen of reference, to except near the offscreen picture that is positioned at the place ahead of this object picture before this object picture, give above-mentioned specific with reference to picture index; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed above-mentioned specific picture with reference to picture index.

According to the present invention, in above-mentioned dynamic image decoding method, above-mentioned specific with reference to picture index, when being decoded, the object picture carries out in the candidate screen of reference, more Zao than this object picture to DISPLAY ORDER, except the candidate screen near the most close offscreen this object picture before this object picture, give; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed above-mentioned specific picture with reference to picture index.

According to the present invention, in above-mentioned dynamic image decoding method, comprise the decoded reference pictures of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And in above-mentioned decoding step, when in direct mode the object piece being decoded, reference is at the first forward direction picture institute's reference of decoded time of said reference piece, that DISPLAY ORDER is more Zao than above-mentioned object picture.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to being positioned near DISPLAY ORDER than the second forward direction picture before the Zao object picture of this object picture, and, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the second forward direction picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), and backward motion vector ((TRB-TRD) * MVR/TRD) of the motion vector of above-mentioned object piece.

According to the present invention, in above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, at the DISPLAY ORDER forward direction picture for the treatment of reference more Zao than above-mentioned object picture, be not present in and keep with reference in the situation in the memory of picture, the picture of, the most close this object picture more Zao than above-mentioned object picture with reference to DISPLAY ORDER.

According to dynamic image decoding method of the present invention, be that a kind of each to a plurality of pictures of consisting of moving image is decoded and will be transformed into corresponding to the bit string of this each picture the dynamic image decoding method of view data, the method comprises the decoding step of object picture to be decoded being decoded with reference to decoded picture; And, above-mentioned decoding step comprises: to as above-mentioned to decode candidate a plurality of with reference to candidate screen of picture of time institute's reference of object picture, the picture that the DISPLAY ORDER of its DISPLAY ORDER and this decoder object picture is nearer, step given in the index with reference to picture index that the value of giving is less, with based on additional in the bit string of above-mentioned object picture, above-mentioned to the object picture encode that the picture of time institute's reference obtains to give with reference to picture index, and above-mentioned index give be endowed in the step with reference to candidate screen with reference to picture index, determine above-mentioned to the object picture decode time institute's reference picture with reference to the picture determining step.

According to dynamic image decoding method of the present invention, that a kind of each to a plurality of pictures of consisting of moving image is decoded respectively, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, the method comprises the decoding step of the object picture that becomes decoder object being decoded with reference at least one decoded picture, in above-mentioned bit string, record and narrate the sign have the above-mentioned object picture of expression when decoded for other later pictures of this object picture, whether to use as the candidate of reference picture, and, in above-mentioned decoding step, based on above-mentioned sign, carry out the management of decoded object picture.

As mentioned above, in the present invention, moving image a kind of a plurality of pictures consisting of moving image are provided each be divided into respectively a plurality of, and by the dynamic image encoding method that each piece is encoded to each picture, the method comprises the coding step that the piece of the object picture that becomes coded object is carried out predictive coding with reference to coded picture; And, in above-mentioned coding step, when coded picture has carried out the B picture of predictive coding of its piece with reference to two at the object picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding, therefore, just can when the B picture is encoded, use the relative immediate forward direction of this B picture with reference to picture.Thus, just can improve the precision of prediction of the motion compensation of B picture, can obtain the raising of code efficiency.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, the B picture that reference has been encoded carries out predictive coding to the piece of this object picture, and be when coded picture has carried out the P picture of predictive coding of its piece with reference to one at the object picture, with reference to any B picture of having encoded each piece of this object picture do not carried out predictive coding.Therefore, just can become with reference to the candidate screen in the memory of the picture of the candidate of picture storage and manage.

In above-mentioned dynamic image encoding method, consist of each of a plurality of pictures of above-mentioned moving image, respectively in above-mentioned coding step as the I picture that does not carry out the coding of its piece with reference to coded picture, with reference to the coded picture P picture that carries out the predictive coding of its piece, and with reference to two the coded picture a certain picture of B picture that carries out the predictive coding of its piece be carried out coding; And, in above-mentioned coding step, when the object picture is the B picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding, and the candidate number with reference to picture of reference when this object picture is encoded as the B picture is equal to or less than the candidate number with reference to picture of reference when this object picture encoded as the P picture.Therefore, just can avoid owing to when the B picture is encoded, causing storing increase with reference to the capacity of the memory of candidate screen with reference to other B picture.

In above-mentioned dynamic image encoding method, consist of each of a plurality of pictures of above-mentioned moving image, respectively in above-mentioned coding step as the I picture that does not carry out the coding of its piece with reference to coded picture, with reference to the coded picture P picture that carries out the predictive coding of its piece, and with reference to two the coded picture a certain picture of B picture that carries out the predictive coding of its piece be carried out coding; And, in above-mentioned coding step, when the object picture is the B picture, to the B picture that the piece of this object picture carries out the reference of predictive coding time institute, just be inserted in this object picture and the relative B picture of this object picture between immediate I on the DISPLAY ORDER or P picture.Therefore, just can improve the precision of prediction in the motion compensation of B picture, and can obtain the raising of code efficiency.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding, and the picture position information B picture of having encoded, that represent the position on the displaying time axle with this reference is included in by in the resulting bit string of the coding of the picture that consists of above-mentioned moving image.Therefore, just can in the decoding side, detect simply when the B picture is encoded employed with reference to candidate B picture as the reference picture.

In above-mentioned dynamic image encoding method, above-mentioned picture position information, nearer by the distance B picture of having encoded from above-mentioned object picture to time institute's reference of encoding at it, on the displaying time axle, code length just shorter code is represented.Therefore, just can reduce the needed size of code of information of being used by the candidate screen of forward direction reference when being illustrated in the decoding side and being identified in the B picture and being encoded.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when the object picture is the B picture, with reference at least one B picture of having encoded the piece of this object picture is carried out predictive coding; And, in by the resulting bit string of the coding of the picture that consists of above-mentioned moving image, comprise as heading message and to be illustrated in when encoding as the B picture of above-mentioned object picture with reference to the information of the B picture situation of having encoded.The B picture that therefore, just can detect simply when the B picture is encoded other in the decoding side is by the situation of forward direction reference.

In addition, in moving image of the present invention, each that a kind of a plurality of pictures consisting of moving image are provided is divided into respectively a plurality of, the dynamic image encoding method of each picture being encoded by each piece, the method comprise with reference at least to one coded picture carry out with reference to and implement its piece predictive coding the P picture and with reference to two coded picture B pictures of carrying out the predictive coding of its piece, the coding step that the object picture that becomes coded object is carried out predictive coding; And, in above-mentioned coding step, be arranged in use this object picture reference pictures of having encoded nearby, be positioned at and this object piece direct mode of the motion vector of the reference block of identical position spatially, when encoding as the object piece of the B picture of object picture, reference is according to the determined coded picture of certain rule, therefore, can make forecasting efficiency optimum according to the coding situation.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to being positioned near the first coded picture before the above-mentioned object picture, that DISPLAY ORDER is more Zao than this object picture.Therefore, the coded prediction efficient of B picture under direct mode is improved.

In above-mentioned dynamic image encoding method, comprise the reference pictures of having encoded of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned coding step, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and this reference block be encoded the second picture of time institute's reference on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the first picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the second picture on the displaying time axle apart from TRB, obtain the forward motion vector (MVR * TRF/TRD) of above-mentioned object piece, and backward motion vector ((TRB-TRD) * MVR/TRD), and use this forward motion vector and this backward motion vector to carry out bi-directional predicted.Therefore, just can be created on well from the motion vector precision of object piece piece in addition the motion vector of the object piece that uses the direct mode.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to the coding forward direction picture the most nearby, that DISPLAY ORDER is more Zao than this object picture that is positioned at above-mentioned object picture, and with reference to being positioned at the backward picture of coding, that DISPLAY ORDER is slower than this object picture of above-mentioned object picture the most nearby, the motion vector that carries out above-mentioned object piece be 0 bi-directional predicted.Therefore, in the process that the B picture is encoded under direct mode, just do not need the calibration of motion vector to process, can obtain the reduction for the treatment of capacity.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, economize the whereabouts prediction error information and be the insertion of bit string image information, corresponding above-mentioned moving image of 0 object piece.Therefore, the reduction of size of code just becomes possibility.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, be 0 o'clock in the prediction error information of above-mentioned object piece, economize the insertion in bit string image information, corresponding above-mentioned moving image of this object piece of whereabouts.Therefore, the reduction of size of code just becomes possibility.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, picture to the candidate that becomes the picture that when the object picture is encoded, carries out reference, give with reference to picture index, and when in direct mode the object piece being encoded, with reference to being endowed specific picture with reference to picture index.Therefore, can make forecasting efficiency optimum according to the coding situation.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, near the picture before the object picture, only as in the situation with reference to the picture of the candidate of picture when the object picture is encoded, when being encoded, the object picture carries out in the candidate screen of reference, to except near the offscreen picture that is positioned at the place ahead of this object picture before this object picture, give above-mentioned specific with reference to picture index.Therefore, the coded prediction efficient of B picture under direct mode is improved and just become possibility.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when being encoded, the object picture carries out in the candidate screen of reference, more Zao than this object picture to DISPLAY ORDER, beyond the picture before this object picture, the candidate screen of the most close this object picture is given above-mentioned specific with reference to picture index.Therefore, the coded prediction efficient of B picture under direct mode is improved and just become possibility.

In above-mentioned dynamic image encoding method, comprising the reference pictures of having encoded of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned coding step, when in direct mode the object piece being encoded, with reference to the first forward direction picture time institute's reference, that DISPLAY ORDER is more Zao than this object picture that is encoded at the said reference piece.Therefore, the coded prediction efficient of B picture under direct mode is improved and just become possibility.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, with reference to being positioned near DISPLAY ORDER than the second forward direction picture before the Zao object picture of this object picture, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the second forward direction picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), and backward motion vector ((TRB-TRD) * MVR/TRD) of the motion vector of above-mentioned object piece.Therefore, just can be created on well from the motion vector precision of object piece piece in addition the motion vector of the object piece that uses the direct mode.

In above-mentioned dynamic image encoding method, in above-mentioned coding step, when in direct mode above-mentioned object piece being encoded, at the DISPLAY ORDER forward direction picture for the treatment of reference more Zao than above-mentioned object picture, be not present in and keep with reference in the situation in the memory of picture, more Zao than above-mentioned object picture with reference to DISPLAY ORDER, the picture of the most close this object picture.Therefore, just can avoid the forecasting efficiency in the process that the B picture is encoded under direct mode greatly to reduce.

In addition, in the present invention, provide a kind of each to a plurality of pictures of consisting of moving image to encode respectively, generation is to the dynamic image encoding method of bit string that should each picture, and the method comprises the coding step of the object picture that becomes coded object being encoded with reference to the picture of having encoded; And above-mentioned coding step comprises, to as above-mentioned to encode candidate a plurality of with reference to candidate screen of picture of time institute's reference of object picture, the picture that the DISPLAY ORDER of its DISPLAY ORDER and this coded object picture is nearer, step given in the index with reference to picture index that the value of giving is less; With will above-mentioned to the object picture encode that the picture of time institute's reference is endowed with reference to picture index, be additional to the indexing step of above-mentioned bit string.Therefore, just can be to as the easiest to be selecteed in time near the picture of object picture with reference to picture, give less of picture index, thus, just can make with reference to the size of code of picture index minimum, and the raising of acquisition code efficiency.

In addition, in the present invention, provide a kind of each to a plurality of pictures of consisting of moving image to encode respectively, generation is to the dynamic image encoding method of bit string that should each picture, and the method comprises that the picture of having processed with reference to having encoded processes the coding step that the object picture of object is encoded to becoming coding; And, in above-mentioned coding step, in above-mentioned bit string, record and narrate the sign whether above-mentioned object picture of expression uses as the candidate of reference picture when processing for the coding of other later pictures of this object picture.Therefore, for example, when being encoded for the B picture of implementing the bi-directional predictive coding processing, the immediate forward direction picture of this B picture position uses as the reference picture and just becomes possibility relatively, just can improve the precision of prediction of the motion compensation of B picture, and can obtain the raising of code efficiency.

In the present invention, provide a kind of respectively to each of a plurality of pictures of consisting of moving image, decode by each piece of processing unit as this picture, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, the method comprises with reference to decoded picture carries out the decoding step of prediction decoding to the piece of object picture to be decoded; And, in above-mentioned decoding step, be when decoded picture has been carried out the B picture of prediction decoding of its piece with reference to two at the object picture, with reference to decoded B picture the piece of this object picture is carried out prediction decoding.Therefore, just the B picture can be used as the candidate screen of forward direction reference and the piece of coded B picture is correctly decoded.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, the decoded B picture of reference carries out prediction decoding to the piece of this object picture, be when decoded picture has been carried out the P picture of prediction decoding of its piece with reference to one at the object picture, with reference to decoded B picture each piece of this object picture do not carried out prediction decoding.Therefore, the management that becomes with reference to the candidate screen in the memory of the picture of the candidate of picture in storage just becomes simple.

In above-mentioned dynamic image decoding method, consist of each of a plurality of pictures of above-mentioned moving image, respectively in above-mentioned decoding step as the I picture that does not carry out the decoding of its piece with reference to decoded picture, with reference to one decoded picture carry out the P picture of the prediction decoding of its piece, and with reference to two the decoded picture a certain picture of B picture that carries out the prediction decoding of its piece be carried out decoding; And, in above-mentioned decoding step, when the object picture is the B picture, the decoded B picture of reference carries out prediction decoding to the piece of this object picture, and the candidate number with reference to picture of reference when this object picture is decoded as the B picture is equal to or less than the candidate number with reference to picture of reference when this object picture decoded as the P picture.Therefore, just can avoid owing to when the B picture is decoded, causing storing increase with reference to the capacity of the memory of candidate screen with reference to other B picture.

In above-mentioned dynamic image decoding method, consist of each of a plurality of pictures of above-mentioned moving image, respectively in above-mentioned decoding step as the I picture that does not carry out the decoding of its piece with reference to decoded picture, with reference to one decoded picture carry out the P picture of the prediction decoding of its piece, and with reference to two the decoded picture a certain picture of B picture that carries out the prediction decoding of its piece be carried out decoding; And, in above-mentioned decoding step, when the object picture is the B picture, the B picture that the piece of this object picture is carried out the reference of prediction decoding time institute, just this object picture and relatively this object picture at the B picture between the immediate I in position or the P picture on the displaying time axle.Therefore, just can improve the precision of prediction in the motion compensation of B picture, and can obtain the raising of code efficiency.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, based on decoded B picture that in above-mentioned bit string, comprise, above-mentioned, be illustrated in the picture position information of the position on the displaying time axle, carry out the piece of this object picture being carried out the processing of prediction decoding with reference to decoded B picture.Therefore, just can in the decoding side, correctly differentiate when the B picture is encoded employed with reference to candidate B picture as the reference picture.

In above-mentioned dynamic image decoding method, above-mentioned picture position information, by the decoded B picture from above-mentioned object picture to time institute's reference of decoding at it, the distance on the displaying time axle is nearer, and code length just shorter code is represented.Therefore, just can subdue the needed size of code of information of when the decoding side is identified in the B picture and is encoded, being used by the candidate screen of forward direction reference.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when the object picture is the B picture, comprise in the bit string based on the picture that consists of above-mentioned moving image in correspondence, be illustrated in when encoding as the B picture of above-mentioned object picture with reference to the heading message of the B picture situation of having encoded, carry out the piece of this object picture being carried out the processing of prediction decoding with reference to decoded B picture.Therefore, just can be in processing for the prediction decoding of object piece, reliably and promptly differentiate when the B picture is encoded other B picture by the situation of forward direction reference.

In addition, in the present invention, provide a kind of respectively to each of a plurality of pictures of consisting of moving image, decode by each piece of processing unit as this picture, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, the method comprises at least reference, to one decoded picture carry out with reference to and implement the P picture of the prediction decoding of its piece, with with reference to two decoded picture carry out the B picture of the prediction decoding of its piece, object picture to be decoded is carried out the decoding step of prediction decoding; And, in above-mentioned decoding step, be arranged in use this object picture decoded reference pictures nearby, be positioned at and this object piece direct mode of the motion vector of the reference block of identical position spatially, when decoding as the object piece of the B picture of object picture, with reference to according to the determined decoded picture of certain rule.Therefore, just can realize corresponding to the coding/decoding method that can make according to the coding situation coding method of forecasting efficiency optimum.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to being positioned near before the above-mentioned object picture, the first coded picture that DISPLAY ORDER is more Zao than this object picture.Therefore, B picture forecasting efficiency in the decoded process under direct mode is improved.

In above-mentioned dynamic image decoding method, comprise the decoded reference pictures of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned decoding step, based on the big or small MVR of the motion vector of said reference piece, above-mentioned backward reference P picture and the second picture of this reference block institute's reference of decoded time on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the first picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the second picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), and backward motion vector ((TRB-TRD) * MVR/TRD) of above-mentioned object piece.Therefore, just can be created on well from the motion vector precision of object piece piece in addition the motion vector of the object piece that uses the direct mode.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to be positioned at above-mentioned object picture the most nearby, the decoding forward direction picture that DISPLAY ORDER is more Zao than this object picture, and with reference to be positioned at above-mentioned object picture the most nearby, the backward picture of decoding that DISPLAY ORDER is slower than this object picture, the motion vector that carries out above-mentioned object piece be 0 bi-directional predicted.Therefore, under direct mode, in the decoded process, just do not need the calibration of motion vector to process at the B picture, can obtain alleviating for the treatment of capacity.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, use the motion vector of said reference piece, to not being included in the above-mentioned bit string, prediction error information is that the image information of 0 object piece is restored.Therefore, just can use the motion vector of other pieces, to be not included in the bit string, prediction error information is that 0 object piece carries out prediction decoding.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, use the motion vector of said reference piece, to not being included in the above-mentioned bit string, prediction error information is that the image information of 0 object piece is restored.Therefore, just can use the motion vector of other pieces, to be not included in the bit string, prediction error information is that 0 object piece carries out prediction decoding.

In above-mentioned dynamic image decoding method, picture to the candidate that becomes the picture that carries out reference when the object picture is decoded is given with reference to picture index, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed specific picture with reference to picture index.Therefore, just can realize corresponding to the coding/decoding method that can make according to the coding situation coding method of forecasting efficiency optimum.

In above-mentioned dynamic image decoding method, near the picture before the object picture, only as in the situation with reference to the picture of candidate when the object picture is decoded, when this is decoded to the object picture, carry out in the candidate screen of reference, to except near the offscreen picture that is positioned at the place ahead of this object picture before this object picture, give above-mentioned specific with reference to picture index; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed above-mentioned specific picture with reference to picture index.Therefore, the decoding forecasting efficiency of B picture under direct mode improved and just become possibility.

In above-mentioned dynamic image decoding method, above-mentioned specific with reference to picture index, when being decoded, the object picture carries out in the candidate screen of reference, more Zao than this object picture to DISPLAY ORDER, beyond the picture before this object picture, the candidate screen of the most close this object picture is given; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference to being endowed above-mentioned specific picture with reference to picture index.Therefore, the decoding forecasting efficiency of B picture under direct mode improved and just become possibility.

In above-mentioned dynamic image decoding method, comprise the decoded reference pictures of said reference piece, be the DISPLAY ORDER backward reference P picture slower than above-mentioned object picture; And, in above-mentioned decoding step, when in direct mode the object piece being decoded, with reference in the institute's reference of decoded time of said reference piece, the first forward direction picture that DISPLAY ORDER is more Zao than above-mentioned object picture.Therefore, the decoding forecasting efficiency of B picture under direct mode improved and just become possibility.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, with reference to being positioned near DISPLAY ORDER than the second forward direction picture before the Zao object picture of this object picture, big or small MVR based on the motion vector of said reference piece, above-mentioned backward reference P picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRD, above-mentioned object picture and above-mentioned the second forward direction picture on the displaying time axle apart from TRF, above-mentioned object picture and above-mentioned the first forward direction picture on the displaying time axle apart from TRB, obtain forward motion vector (MVR * TRF/TRD), and backward motion vector ((TRB-TRD) * MVR/TRD) of the motion vector of above-mentioned object piece.Therefore, just can be created on well from the motion vector precision of object piece piece in addition the motion vector of the object piece that uses the direct mode.

In above-mentioned dynamic image decoding method, in above-mentioned decoding step, when in direct mode above-mentioned object piece being decoded, at the DISPLAY ORDER forward direction picture for the treatment of reference more Zao than above-mentioned object picture, be not present in and keep with reference in the situation in the memory of picture, more Zao than above-mentioned object picture with reference to DISPLAY ORDER, the picture of the most close this object picture.Therefore, just can avoid B picture forecasting efficiency in the decoded process under direct mode to reduce in a large number.

In addition, in the present invention, provide a kind of each to a plurality of pictures of consisting of moving image to decode respectively, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, the method comprises the decoding step of object picture to be decoded being decoded with reference to decoded picture; And above-mentioned decoding step comprises, to as above-mentioned to decode candidate a plurality of with reference to candidate screen of picture of time institute's reference of object picture, the picture that the DISPLAY ORDER of its DISPLAY ORDER and this decoder object picture is nearer, step given in the index with reference to picture index that the value of giving is less; With based on additional in the bit string of above-mentioned object picture, above-mentioned to the object picture encode that the picture of time institute's reference is endowed with reference to picture index, and above-mentioned index give be endowed in the step with reference to candidate screen with reference to picture index, determine above-mentioned to the object picture decode time institute's reference picture with reference to the picture determining step.Therefore, just can correctly decode, by can to the easiest selecteed in time near the picture of object picture, give the bit string that the less high coding method of the code efficiency with reference to picture index generates.

In addition, in the present invention, provide a kind of each to a plurality of pictures of consisting of moving image to decode respectively, to be transformed into to bit string that should each picture the dynamic image decoding method of view data, the method comprises that the picture of having processed with reference to having decoded processes the decoding step that the object picture of object is decoded to becoming decoding; In above-mentioned bit string, record and narrate the sign have the above-mentioned object picture of expression when processing for the decoding of other later pictures of this object picture, whether to use as the candidate of reference picture, and in above-mentioned decoding step, based on above-mentioned sign, implemented the management of the object picture of decoding processing.Therefore, just can correctly decode,,, except the P picture of having implemented the forward predictive coded processing, also use and implemented the B picture bit string that generate, corresponding B picture that bi-directional predictive coding is processed with reference to picture as forward direction.

Description of drawings

Fig. 1 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 1.

Fig. 2 (a) and 2 (b) are the schematic diagrames that explanation utilizes the dynamic image encoding method of embodiments of the present invention 1, the wherein arrangement of the DISPLAY ORDER of Fig. 2 (a) expression picture, the arrangement of the coded sequence of Fig. 2 (b) expression picture.

Fig. 3 is the schematic diagram of the moving image decoding apparatus of the explanation dynamic image encoding device of execution mode 1 and execution mode 2, represents the P in the diode-capacitor storage in the lump, the method for B picture.

Fig. 4 (a) and 4 (b) are the key diagrams of execution mode 1, the 1st example (Fig. 4 (a)) and the 2nd example (Fig. 4 (b)) of expression direct mode coding (picture B11).

Fig. 5 (a) and 5 (b) are the schematic diagrames of execution mode 1, the 3rd example (5 (a)) and the 4th example (5 (b)) of expression direct mode coding (picture B11).

Fig. 6 (a)-6 (c) is the key diagram of execution mode 1, and the 5th example (6 (a)) of expression direct mode coding (picture B11) is jumped over piece (Fig. 6 (b)) and jumped over identifier (Fig. 6 (c)).

Fig. 7 (a) and 7 (b) are the key diagrams of execution mode 1, the 1st example (7 (a)) and the 2nd example (7 (b)) of expression direct mode coding (picture B12).

Fig. 8 (a) and 8 (b) are the key diagrams of execution mode 1, the 3rd example (8 (a)) and the 4th example (8 (b)) of expression direct mode coding (picture B12).

Fig. 9 (a) and 9 (b) are the key diagrams of execution mode 1, and expression is with reference to the 1st example (9 (a)) of the B picture more on the front of immediate forward direction P picture and the 2nd example (, 9 (b)).

Figure 10 (a) and 10 (b) are the key diagrams of execution mode 1, and expression is not with reference to the 1st example (10 (a)) and the 2nd example (10 (b)) of the B picture more on the front of immediate forward direction I picture or P picture.

Figure 11 is the key diagram of execution mode 1 and 2, the 1st method that expression separately manages the P picture in the memory and B picture.

Figure 12 is the key diagram of execution mode 1 and 2, the 2nd method that expression separately manages the P picture in the memory and B picture.

Figure 13 is the key diagram of execution mode 1 and 2, the 3rd method that expression separately manages the P picture in the memory and B picture.

Figure 14 is the key diagram of execution mode 1 and 2, the 4th method that expression separately manages the P picture in the memory and B picture.

Figure 15 is the block diagram that explanation utilizes the moving image decoding apparatus of embodiments of the present invention 2 to use.

Figure 16 (a) and 16 (b) are the schematic diagrames that explanation utilizes the dynamic image decoding method of embodiments of the present invention 2, the wherein arrangement of the decoding order of 16 (a) expression picture, the arrangement of the DISPLAY ORDER of 16 (b) expression picture.

Figure 17 is the key diagram of execution mode 2, represents bi-directional predicted decoding (picture B11).

Figure 18 (a) and 18 (b) are the key diagrams of execution mode 2, the 1st example (18 (a)) and the 2nd example (18 (b)) of expression direct mode decoding (picture B11).

Figure 19 (a) and 19 (b) are the key diagrams of execution mode 2, the 3rd example (19 (a)) and the 4th example (19 (b)) of expression direct mode decoding (picture B11).

Figure 20 is the key diagram of execution mode 2, represents bi-directional predicted decoding (picture B12).

Figure 21 (a) and 21 (b) are the key diagrams of execution mode 2, the 1st example (21 (a)) and the 2nd example (21 (b)) of expression direct mode decoding (picture B12).

Figure 22 (a) and 22 (b) are the key diagrams of execution mode 2, the 3rd example (22 (a)) and the 4th example (22 (b)) of expression direct mode decoding (picture B12).

Figure 23 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 3 to use.

Figure 24 is the schematic diagram of dynamic image encoding device of explanation execution mode 3, represents the P in the diode-capacitor storage in the lump, the method for B picture.

Figure 25 (a) and 25 (b) are the key diagrams of above-mentioned execution mode 3, and the situation (25 (a)) near the decoding of the B picture after the P picture is not carried out in expression, and the situation of the decoding of the picture of not being scheduled to.

Figure 26 is the block diagram that explanation utilizes the moving image decoding apparatus of embodiments of the present invention 4.

Figure 27 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 5.

Figure 28 is the key diagram of execution mode 5, the management method of expression picture memory and with reference to the adding method of picture index.

Figure 29 (a) and 29 (b) are the key diagrams of execution mode 5, and the picture of expression DISPLAY ORDER is arranged the picture of (29 (a)) and coded sequence and arranged.

Figure 30 is the key diagram of execution mode 5, the management method of expression picture memory and with reference to the adding method of picture index.

Figure 31 is the key diagram of execution mode 5, expression use two systems with reference in the situation of picture index, the data structure of the bit string of corresponding blocks.

Figure 32 is the block diagram that explanation utilizes the moving image decoding apparatus of embodiments of the present invention 6 to use.

Figure 33 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 7.

Figure 34 (a) and 34 (b) are the schematic diagrames that explanation utilizes the dynamic image encoding method of embodiments of the present invention 7, the arrangement (34 (b)) of the coded sequence of the arrangement (34 (a)) of the DISPLAY ORDER of expression picture and expression picture.

Figure 35 is the block diagram that explanation utilizes the moving image decoding apparatus of embodiments of the present invention 8.

Figure 36 (a) and 36 (b) are the schematic diagrames that explanation utilizes the dynamic image encoding method of embodiments of the present invention 7, the arrangement (36 (b)) of the DISPLAY ORDER of the arrangement (36 (a)) of the decoding order of expression picture and expression picture.

Figure 37 is the key diagram of execution mode 8, the management method of expression picture memory.

To be expression realize the storage medium of program of the device of each execution mode by software for Figure 38 (a) and 38 (b), and Figure 38 (c) is the figure that the computer system of storage medium is used in expression.

Figure 39 is the figure of the application example of the explanation dynamic image encoding method of the respective embodiments described above and dynamic image decoding method, and expression realizes the content provider system of content converting service.

Figure 40 is the figure that explanation has utilized the portable phone of the dynamic image encoding method of the respective embodiments described above and dynamic image decoding method.

Figure 41 is the block diagram of the detailed structure of expression portable phone shown in Figure 40.

Figure 42 is that expression has utilized the digital broadcasting of the dynamic image encoding device of the respective embodiments described above and moving image decoding apparatus with the concept map of system.

Figure 43 (a) and 43 (c) are the key diagrams of existing dynamic image encoding method, expression consists of the arrangement (43 (a)) of the picture of moving image, distinguishes picture bar (43 (b)) and the macro block (43 (c)) of picture.

Figure 44 is the figure of the coded data of the general moving image of explanation, and expression is to the encode structure of resulting stream of each picture that consists of moving image.

Figure 45 is the figure of the moving image encoding mode of explanation existing MPEG etc., the indicated object picture with at this to the encode relation of picture of time institute's reference of object picture.

Figure 46 (a) and 46 (b) they are the key diagrams of existing direct mode coding, are illustrated in the motion vector (46 (a)) that uses in the direct mode, and for the relative position (46 (b)) of the piece of picture.

Figure 47 (a) and 47 (b) are the key diagrams of existing adding method with reference to picture index, and expression is given for the index that carries out the candidate screen of reference at the P picture, when the B picture is encoded.

Figure 48 (a) and 48 (b) are the key diagrams of existing dynamic image encoding method, the arrangement (48 (a)) of the DISPLAY ORDER of the picture of expression formation moving image, and the arrangement of coded sequence (48 (b)).

Figure 49 is the key diagram of existing dynamic image encoding method, and expression is for the management example with reference to picture memory of the picture of the arrangement of above-mentioned coded sequence.

Figure 50 (a) and 50 (b) are the figure of the problem points of the existing inter-picture prediction coding method of expression, are illustrated in the situation (50 (a)) of carrying out two-way reference and the situation (50 (b)) of two pictures being carried out the rear reference.

Figure 51 (a) and 51 (b) are the figure of the problem points of the existing adding method with reference to picture index of expression, the arrangement (51 (b)) of the arrangement (51 (a)) of the picture of expression DISPLAY ORDER and the picture of coded sequence.

Embodiment

[execution mode 1]

Fig. 1 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 1 to use.

The dynamic image encoding device 10 of present embodiment 1 is divided into respectively fixing data to each of a plurality of pictures that consist of moving image and processes unit (piece), and by each piece the view data of each picture encoded.

In specific words, this dynamic image encoding device 10 has view data (input data) Id of the picture that storage inputs, the input picture memory 101 of the data I d that stores by each piece output simultaneously; Will be from above-mentioned with reference to picture memory 101 output, become the differential data of the prediction data Pd of the view data Md of object piece of coded object and this object piece, the calculus of differences unit 102 that calculates as the prediction error data PEd of object piece; The coded prediction error unit 103 that carries out compressed encoding with view data Md or prediction error data PEd to above-mentioned object piece.Above-mentioned with reference to picture memory 101 in, based on object picture and relation to the picture (with reference to picture) of its predictive coding time institute reference, carry out the processing that coded sequence by picture changes the view data of the picture that row inputs by DISPLAY ORDER.

Dynamic image encoding device 10, output data (coded data) Ed that also has above-mentioned coded prediction error unit 103 expands decoding, and the predicated error decoding unit 105 of the decoding differential data PDd of object output piece; The decoding differential data PDd of this object piece of addition and the prediction data Pd of above-mentioned object piece, and the add operation unit 106 of the decoded data Dd of object output piece; With record this decoded data Dd based on memory control signal Cd2, and with the decoded data Dd that stores as the data Rd in the candidate (candidate screen) of the picture of the coding time institute reference of carrying out the object piece export with reference to picture memory 117.

Dynamic image encoding device 10, also have based on above-mentioned output data with reference to picture memory 101 (view data of object piece) Md with reference to output data (data of the candidate screen) Rd of picture memory 117, the motion vector detecting unit 108 of the motion vector MV of detected object piece; With based on the motion vector MV of this object piece and above-mentioned each with reference to picture memory 101,117 output data M d, Rd, determine the coded system of corresponding objects piece, simultaneously the mode selected cell 109 of output switch control signal Cs2.Above-mentioned motion vector detecting unit 108 with reference to a plurality of candidate screen that can reference when the predictive coding of object piece, detects the motion detection of above-mentioned motion vector.In addition, aforesaid way selected cell 109 to above-mentioned object piece, selects a coded system so that code efficiency becomes optimum from a plurality of coded systems.At this moment, under inter-picture prediction is encoded selecteed situation, from selecting best picture a plurality of candidate screen of reference.

In the dynamic image encoding device 10 of this execution mode 1, for the P picture (namely, when a piece is carried out predictive coding, with reference to the picture of a picture of having encoded), select coded system in the picture, use the inter-picture prediction coded system of motion vector and do not use in the inter-picture prediction coded system of motion vector (namely motion vector is 0) one.In addition, for the B picture (namely, when a piece is carried out predictive coding, picture with reference to two pictures of having encoded), in coded system, the inter-picture prediction coded system of using forward motion vector, the inter-picture prediction coded system of using backward motion vector, the inter-picture prediction coded system of using bi-directional motion vector and the direct mode in the selection picture.In addition, in this execution mode 1, when in direct mode the piece of above-mentioned B picture being encoded, with reference on the displaying time axle near the coded picture before the object picture.

In addition, dynamic image encoding device 10, have be arranged on above-mentioned with reference to the selector switch 111 between picture memory 101 and the calculus of differences unit 102; Be arranged on the selector switch 112 between calculus of differences unit 102 and the coded prediction error unit 103; Be arranged on reference to the connection cut-off switch 113 between picture memory 101 and aforesaid way selected cell 109 and the motion vector detecting unit 108; Be arranged on the connection cut-off switch 114 between aforesaid way selected cell 109 and the add operation unit 106 and be arranged on connection cut-off switch 115 between above-mentioned coded prediction error unit 103 and the predicated error decoding unit 105.

In addition, dynamic image encoding device 10, the connection that has to control by switch controlling signal Cs1 above-mentioned switch 113～115 disconnects action, and output code generates the coding-control unit 110 of control signal Cd1 and memory control signal Cd2 simultaneously; And generate control signal Cd1 based on code output data (coded data) Ed of coded prediction error unit 103 is carried out Variable Length Code, and output is corresponding to the bit string generation unit 104 of the bit string Bs of object piece.In this bit string generation unit 104, input has motion vector that expression detects by motion vector detecting unit 108 and by the information of mode selected cell 109 determined coded system Ms.In the bit string Bs corresponding to this object piece, include expression corresponding to the information of motion vector MV and the coded system Ms of object piece.

Above-mentioned selector switch 111 have an input terminal Ta and two lead-out terminal Tb1 and Tb2, and according to above-mentioned switch controlling signal Cs2, this input terminal Ta is connected to the side of above-mentioned two lead-out terminal Tb1, Tb2.Above-mentioned selector switch 112 has two input terminal Tc1 and Tc2 and a lead-out terminal Td, and according to above-mentioned switch controlling signal Cs2, this lead-out terminal Td is connected to the side of above-mentioned two input terminal Tc1, Tc2.In addition, by above-mentioned selector switch 111, to the view data Md of input terminal Ta input from 101 outputs of reference picture memory, from a side lead-out terminal Tb1 to the input terminal Tc1 output of selector switch 112 from above-mentioned view data Md with reference to picture memory 101, from the opposing party's lead-out terminal Tb2 to 102 outputs of calculus of differences unit from above-mentioned view data Md with reference to picture memory 101.By above-mentioned selector switch 112, input terminal Tc1 to a side inputs from above-mentioned view data Md with reference to picture memory 101, input terminal Tc2 to the opposing party inputs by calculus of differences unit 102 resulting differential data PEd, from lead-out terminal Td, the side of this view data Md and differential data PEd is output to coded prediction error unit 103.

The below describes the action of dynamic image encoding device 10.

In the following description, displaying time wants Zao picture (forward direction picture) to refer to that relative object picture is positioned at the picture in the place ahead in time than the picture that becomes coded object (object picture), perhaps also is called simply the picture in the place ahead that is positioned at the object picture.In addition, the displaying time picture (backward picture) slower than object picture refers to that relative object picture is positioned at the picture at rear in time, perhaps also is called the picture at the rear that is positioned at the object picture simply.In addition, in the explanation of each execution mode below, the expression that " on the time " is such is as long as no the implication of doing in advance special instruction and then represent " pressing the displaying time order ".

In the dynamic image encoding device 10 of execution mode 1, input image data Id is imported into reference to picture memory 101 with picture unit by the displaying time order.

Fig. 2 (a) is the figure of explanation stored order of view data of each picture in reference picture memory 101.In Fig. 2 (a), ordinate represents picture.Symbol shown in the bottom right of each picture, first letter represent picture type (I, P or B), the picture index of the numeral time sequencing of back.Namely, picture P1, B2, B3, P4, B5, B6, P7, B8, B9, P10, B11, B12, P13, B14, B15, P16 shown in Fig. 2 (a), be equivalent to picture F (k+3) shown in Figure 45～F (k+18) [k=-2], and these pictures, press DISPLAY ORDER, namely the beginning of morning is arranged in order along displaying time axle X from DISPLAY ORDER.

In reference picture memory 101, the view data of picture is pressed the DISPLAY ORDER storage of picture.In the view data of this picture of storing in reference to picture memory 101, changed the order of lining up corresponding to the coding of picture from the DISPLAY ORDER of picture.Below, for the purpose of simplifying the description, the view data of picture also is called picture simply.

Specifically, the order with picture in reference picture memory 101 changes the processing of lining up coded sequence from input sequence (DISPLAY ORDER), carries out based on the reference relation in the inter-picture prediction coding.Namely, carry out this and change row's processing, make when the first picture is encoded before the first picture, to be encoded as employed the second picture of reference picture.

When the P picture coding, relative coding object picture (P picture) is in nearby 3 pictures (I or P picture) in the place ahead in time, is used as the candidate screen of reference picture.In the predictive coding of the piece of P picture, from above-mentioned 3 candidate screen with reference to maximum 1.

In addition, when the B picture is encoded, relatively object picture (B picture) be positioned in time 2 of the place ahead near picture (I or P picture), relatively the object picture be positioned in time the immediate B picture in the place ahead and relatively the object picture be positioned in time I or the P picture at rear, use as the candidate screen of reference picture.In the predictive coding of the piece of B picture, from above-mentioned 4 candidate screen with reference to maximum 2.

Specifically, picture P10, B11, B12, P13, and to the corresponding relation with reference to the candidate screen of picture that should each picture, shown in the arrow among Fig. 2 (a).Namely, when P picture P10 was encoded, picture P1, P4, P7 were used as the candidate screen with reference to picture, and when P picture P13 was encoded, picture P4, P7, P10 were used as the candidate screen with reference to picture.In addition, when B picture B11 was encoded, picture P7, B9, P10, P13 were as the candidate screen with reference to picture, and when B picture B12 was encoded, picture P7, P10, B11, P13 were as the candidate screen with reference to picture.

In addition, Fig. 2 (b) expression will be from the order of the picture of the picture P1～P16 shown in Fig. 2 (a), changes the order of lining up the picture behind the coded sequence from DISPLAY ORDER.After changing row, picture shown in Fig. 2 (a), shown in Fig. 2 (b), on time shaft (scramble time axle) Y of presentation code time, according to its scramble time order, namely arrange by the order of picture P4, B2, B3, P7, B5, B6, P10, B8, B9, P13, B11, B12, P16.

In reference picture memory 101, carried out changing the data of row's picture, according to the scramble time, each predetermined data has been processed unit, called over.In this execution mode, it is the data unit of carrying out motion compensation that data are processed unit, more specifically, is exactly the rectangular-shaped image space (macro block) of all having arranged in the horizontal direction 16 pixels with vertical direction.But in the following description, macro block also is called piece simply.

Below, the coding of picture P13, B11, B12 processed describing in order.

The coding of＜picture P13 is processed 〉

At first, the coding processing of picture P13 described.

Because becoming picture (object picture) P13 of coded object is the P picture, so for the inter-picture prediction of the object piece among picture P13 coding, be to carry out being positioned in time the place ahead or rear 1 the unidirectional inter-picture prediction coding of coded picture with reference to relative object picture.

In the following description, the P picture that just uses the place ahead that is positioned at the object picture describes as the situation of reference picture.

In the case, as the coding of picture P13, use the inter-picture prediction coding of forward direction reference.In addition, in the process that the P picture is encoded, do not use the B picture as the reference picture.Thereby, in the case, with reference to the candidate screen of picture, be 3 pictures (I or P picture) in the place ahead, specifically, be exactly picture P4, P7, P10.The coding of these candidate screen has finished when the object picture begins to encode, and the data of corresponding candidate screen (decoded data) Dd is stored in reference to picture memory 117 with reference in the picture memory 101.

In the process that the P picture is encoded, each switch is controlled by switch controlling signal Cs1 in coding-control unit 110, so that switch 113,114 and 115 becomes on-state.The view data Md of the macro block of the corresponding picture P13 that reads from reference picture memory 101 is input to motion vector detecting unit 108, mode selected cell 109 and calculus of differences unit 102.

Motion vector detecting unit 108 is used to be stored in reference to picture memory with reference to picture P4, P7 in the picture memory 117, the decode image data Rd of P10, detects the motion vector MV of the macro block of picture P13.Then, detected motion vector MV is output to mode selected cell 109.

Mode selected cell 109 uses by the detected motion vector of motion vector detecting unit 108, determines the coded system of the piece of picture P13.Coded system represents with what kind of method piece to be encoded.For example, as mentioned above, in the situation of P picture, the inter-picture prediction of coding, use motion vector is encoded, is not used the inter-picture prediction coding of motion vector (motion is made as 0 to be processed) in picture, determines a coded system.Determining when determining coded system, generally is to select to make to give as size of code in the quantity of the position that will be scheduled to that encoding error is minimum coded system in the situation of piece.At this moment, in the situation of selecting the inter-picture prediction coding, from picture P4, P7, P10, select optimal one as the reference picture.

By mode selected cell 109 determined coded system Ms, be output to bit string generation unit 104.In addition, under determined coded system Ms was situation with reference to the coded system of forward direction picture, the resulting vector of motion detection (forward motion vector) MVp by reference forward direction picture and this motion vector were that the information Rp of which detected motion vector with reference to picture P4, P7, P10 and so on also is output to bit string generation unit 104.

Represented in the situation of inter-picture prediction coded system by mode selected cell 109 determined coded system Ms, employed motion vector MVp and this motion vector are the information Rp of which the detected motion vector and so on reference to picture P4, P7, P10 in inter-picture prediction coding, are stored in motion vector storage unit 116.

In addition, in mode selected cell 109, and for the corresponding motion compensation of the determined coded system of object piece, use with reference to picture with for the motion vector of object piece and carry out.So resulting by this motion compensation, for the prediction data Pd of object piece, be output to calculus of differences unit 102 and add operation unit 106.

But, selecting in the picture in the situation of coded system generation forecast data Pd not in mode selected cell 109.In addition, make input terminal Ta be connected to lead-out terminal Tb1 by control switch 111, and make lead-out terminal Td be connected to input terminal Tc1 by control switch 112 in the situation of coded system having selected in the picture.On the other hand, in the situation of having selected the inter-picture prediction coding, make input terminal Ta be connected to lead-out terminal Tb2 by control switch 111, and make lead-out terminal Td be connected to input terminal Tc2 by control switch 112.

Below, just to selecting the inter-picture prediction coded system to describe as the situation of coded system Ms by mode selected cell 109.

The view data Md of the piece of input picture P13 and from the prediction data Pd of the correspondence of mode selected cell 109 in calculus of differences unit 102.Calculus of differences unit 102 calculates the view data Md of piece of picture P13 and the differential data of prediction data Pd, and this differential data is exported as prediction error data PEd.

Prediction error data PEd is imported into coded prediction error unit 103.Coded prediction error unit 103 is implemented the codings such as frequency translation and quantification to the prediction error data PEd that inputs and is processed and generation coded data Ed.These processing, such as processing such as frequency translation and quantifications, take correspondence in the horizontal direction with vertical direction on all arranged the rectangular-shaped image space (sub-block) of 8 pixels data carry out as unit.

The coded data Ed of 103 outputs is imported into bit string generation unit 104 and predicated error decoding unit 105 from the coded prediction error unit.

Bit string generation unit 104 carries out Variable Length Code to the coded data Ed that inputs and generates bit string.And then, the information of the Ms such as additional movement vector MVp and coded system in this bit string, and from heading message of coding-control unit 110 etc. and generate bit string Bs.

Be to carry out in the situation of coded system of the place ahead reference at coded system Ms, information (with reference to the image information) Rp which of picture P4, P7, P10 to detect motion vector and so on reference to also is affixed in the bit string.

Then, to the management method of reference picture memory, and the information with reference to picture (with reference to image information) in the expression candidate screen describes.

Fig. 3 is illustrated in the situation that changes along with the process of time with reference to the picture of storing in the picture memory 117.This carries out according to the memory control signal Cd2 from coding-control unit 110 with reference to the management of picture memory 117 with reference to picture memory 117.In addition, the storage area (#1)～(#5) that has 5 picture amounts with reference to picture memory 117.In each storage area, can store the view data of 1 picture amount.But the memory cell of each picture is not limited to 1 with reference to the zone in the picture memory, also can be a memory.

At first, the management method of reference picture memory described.

When the coding of picture P13 begins, in each storage area (#1)～(#5) of reference picture memory 117, maintain picture B8, P4, P7, P10, B9.Although picture B9 does not use in the process that picture P13 is encoded, in order to use in the process that is encoded at picture B11, and remain in reference in the picture memory 117.Picture P13 with picture P4, P7 and P10 as encoding with reference to the candidate screen of picture.Coded picture P13 is held in the storage area (#1) that has kept picture B8.It is the reasons are as follows.Although picture P4, P7, P10 and B9 are used as the candidate screen with reference to picture when picture P13 and later picture thereof are encoded, picture B8 can not be used as with reference to picture when these picture codings.In Fig. 3, the picture shown in the circular mark is the moment of having finished at the coding of object picture, is stored at last the picture (object picture) with reference to picture memory 117.

Then, for the concrete method of giving each candidate screen as the information of reference picture with reference to picture index is explained.

This is with reference to picture index, is to be illustrated in when each piece encoded, and which picture is used as the information with reference to picture in the candidate screen of reference picture.In other words, above-mentioned with reference to picture index, be to use picture P4, the P7 of candidate, which the information among the P10 when being illustrated in the motion vector of object piece of detected object picture (picture P13).In the giving of reference picture index, use from distance object picture in time nearest candidate screen begin to give in order the method for index.

Specifically be exactly, when the piece of object picture P13 is encoded, picture P10 is designated as in the situation with reference to picture, and then the candidate screen in tight the place ahead of additional representation object picture P13 is designated as information (with reference to picture index [0]) with reference to picture in the bit string of object picture P13.When the piece of object picture P13 was encoded, in the situation that picture P7 is referenced, then the candidate screen before 2 of additional representation object picture P13 was designated as information (with reference to picture index [1]) with reference to picture in the bit string of object picture P13.When the piece of object picture P13 was encoded, in the situation that picture P4 is referenced, then the candidate screen before 3 of additional representation object picture P13 was designated as information (with reference to picture index [2]) with reference to picture in the bit string of object picture P13.

In Fig. 3, represent, be endowed the picture of symbol [b] as the reference image information, after when the object picture is encoded, will becoming to the candidate of reference picture.

The coding of＜picture B11 is processed 〉

The below processes the coding of picture B11 and describes.

Because picture (object picture) to be encoded is picture B11, so for the inter-picture prediction of the object piece among picture B11 coding, carry out being positioned in time the place ahead or rear 2 predictive codings between the bi-directional picture of coded picture with reference to relative object picture.

Below, just as with reference to picture, the situation of using a picture (I picture, P picture or B picture) in the place ahead that is positioned at the object picture and being positioned at a picture (I picture or P picture) at the rear of object picture describes.

Namely, in the case, as the candidate screen of forward direction with reference to picture, two pictures (I or P picture) that use relative object picture (picture B11) to approach in time, and the relative object picture immediate B picture in position in time.In addition, with reference to picture, use relative object picture to be in time immediate I or P picture as backward.Thereby, in the case, for the candidate screen with reference to picture of picture B11, be exactly picture P7, B9 and P10 (forward direction picture) and picture P13 (backward picture).

When preparation was used as encoding with reference to the B picture of picture when other pictures are encoded, coding-control unit 110 was controlled each switch by switch controlling signal Cs1 and is made switch 113,114,115 become on-state.Because picture B11 is used as with reference to picture when other pictures are encoded, so coding-control unit 110 is controlled each switch by switch controlling signal Cs1 and made switch 113,114,115 become on-state.Read from reference picture memory 101, the view data Md of the piece of corresponding picture B11 is imported into motion vector detecting unit 108, mode selected cell 109 and calculus of differences unit 102.

Motion vector detecting unit 108 is carried out the detection of forward motion vector and backward motion vector for the object piece of picture B11.When these motion vector detection, be stored in reference to picture P7, B9 and P10 in the picture memory 117, be used as forward direction with reference to the candidate screen of picture, picture P13 is used as backward with reference to picture.The detection of backward motion vector is based on carrying out as backward picture P13 with reference to picture.By the detected motion vector of motion vector detecting unit 108, be output to mode selected cell 109.

Mode selected cell 109 uses by the detected motion vector of motion vector detecting unit 108, determines the coded system of the object piece of picture B11.In this occasion, the coded system of B picture B11, for example, in the picture coding, used forward motion vector the inter-picture prediction coding, used backward motion vector the inter-picture prediction coding, used the inter-picture prediction coding of bi-directional motion vector, and select in the direct mode.Be to have used in the situation of inter-picture prediction coding of forward motion vector in coded system, from picture P7, B9 and P10, select optimal picture as the reference picture.

Below, the processing of in direct mode piece of picture B11 being encoded is described.

[the 1st example of direct mode coding]

The 1st example that Fig. 4 (a) expression is encoded to piece (object piece) BLa1 of picture (object picture) B11 in direct mode.The conduct that this direct mode coding utilizes the rear that is in picture B11 is with reference to motion vector (benchmark motion vector) MVc1 with piece (reference block) BLb1 of object piece BLa1 same position of being among picture (reference pictures) P13 of picture.Motion vector MVc1 is the employed motion vector when encoding of the piece BLb1 to picture P13, and is stored in the motion vector storage unit 116.This motion vector MVc1 is detected with reference to above-mentioned picture P10, and the regional CRc1 of the corresponding blocks BLb1 of expression picture P10.To piece BLa1, use motion vector MVd1, the MVe1 parallel with motion vector MVc1, and as with reference to the selected picture P10 of picture, P13, implement bi-directional predictive coding and process.The motion vector MVd1 that when this piece BLa1 is encoded, uses, and MVe1 be respectively the forward motion vector of regional CRd1 of the corresponding blocks BLa1 of expression picture P10, and the backward motion vector of the regional CRe1 of the corresponding blocks BLa1 of expression picture P13.

In the case, the big or small MVF of forward motion vector MVd1 and the big or small MVB of backward motion vector MVe1 can be obtained respectively by formula (1), (2).

MVF＝MVR×TRF/TRD ...(1)

MVB=(TRB-TRD) * MVR/TRD ... (2) wherein, MVF, MVB be horizontal component and the vertical component of expressive movement vector respectively.

In addition, MVR is the size (by the direction on sign symbol (sign) expression 2 dimension spaces) of motion vector MVc1, TRD be object picture (picture B11) backward with reference to picture (picture P13) and after this to the be encoded time gap of picture (picture P10) of time institute's reference of piece BLb1 of reference picture (picture P13).In addition, TRF is object picture (picture B11) and near the time gap with reference to picture (picture P10) before it, and TRB is object picture (picture B11) and in the be encoded time gap of picture (picture P10) of time institute's reference of rear piece BLb1 to the reference picture.

[the 2nd example of direct mode coding]

The 2nd example of the coding processing of using direct mode then, is described.

The 2nd example that Fig. 4 (b) expression is encoded to piece (object piece) BLa2 of picture (object picture) B11 in direct mode.

The conduct that this direct mode coding utilizes the rear that is in picture B11 is with reference to motion vector (benchmark motion vector) MVf2 with piece (reference block) BLb2 of piece BLa2 same position of being among the picture P13 of picture.Motion vector MVf2 is piece BLb2 employed motion vector when being encoded, and is stored in the motion vector storage unit 116.This motion vector MVf2 is detected with reference to above-mentioned picture P7, and the regional CRf2 of the corresponding blocks BLb2 of expression picture P7.To piece BLa2, use motion vector MVg2, the MVh2 parallel with motion vector MVf2, and as with reference to the selected picture P10 of picture and picture P13, implement bi-directional predictive coding and process.The motion vector MVg2 that when this piece BLa2 is encoded, uses, and MVh2 represent respectively the forward motion vector of regional CRg2 of the corresponding blocks BLa2 of picture P10 and the backward motion vector of the regional CRh2 of the corresponding blocks BLa2 of expression picture P13.

In the case, motion vector MVg2, and big or small MVF, the MVB of MVh2 can be obtained by above-mentioned formula (1), (2) respectively.

As mentioned above, in direct mode, to when object piece BLa2 is encoded as rear in the reference picture, be positioned at this object piece relatively the motion vector MVf2 of the BLb2 of identical position calibrate, thereby obtain forward motion vector MVg2 and backward motion vector MVh2 for the object piece.So, in the selecteed situation of direct mode, just do not need to send the information of the motion vector of object piece.And, since as forward direction with reference to picture, use apart from the object picture immediate coded picture in position in time, so forecasting efficiency is improved.

[the 3rd example of direct mode coding]

The 3rd example of the coding processing of using direct mode then, is described.

The 3rd example that Fig. 5 (a) expression is encoded and processed piece (object piece) BLa3 of picture (object picture) B11 in direct mode.

This direct mode coding utilize as among backward picture with reference to picture (reference pictures) P13 of picture B11, be in motion vector (benchmark motion vector) MVc3 with the piece BLb3 of piece BLa3 same position.Motion vector MVc3 is employed motion vector when piece BLb3 is encoded, and is stored in the motion vector storage unit 116.This motion vector MVc3 is detected with reference to above-mentioned picture P7, and the regional CRc3 of the corresponding blocks BLb3 of expression picture P7.To piece BLa3, use motion vector MVd3, the MVe3 parallel with motion vector MVc3, with the picture of the time institute's reference that is encoded at piece BLb3 (as forward direction with reference to the selected picture P7 of picture) and as backward picture P13 with reference to picture, implement bi-directional predicted processing.In the case, motion vector MVd3, the MVe3 that uses when object piece BLa3 is encoded represents respectively the forward motion vector of regional CRd3 of the corresponding blocks BLa3 of picture P7, and the backward motion vector of the regional CRe3 of the corresponding blocks BLa3 of expression picture P13.

In the case, big or small MVF, the MVB of motion vector MVd3, MVe3 can be obtained by following formula (3) and top formula (2) respectively.

MVF＝MVR×TRB/TRD...(3)

Wherein, MVR is the size of motion vector MVc3.

As mentioned above, in the coding of the direct mode shown in Fig. 5 (a), to when the object piece is encoded as rear in the reference picture, be positioned at this object piece relatively the motion vector MVc3 of the piece BLb3 of identical position calibrate, thereby obtain forward motion vector MVd3 and backward motion vector MVe3 for the object piece, so in the selecteed situation of direct mode, just do not need to send the information of the motion vector of object piece.

When the picture P13 of the time institute's reference that is encoded at piece BLb3 from reference picture memory 117 in the situation of deletion, immediate forward direction is in time got final product with reference to picture as the forward direction in the direct mode with reference to picture P10.Processing in the direct mode in the case, identical with the processing (the 1st example) of the direct mode shown in Fig. 4 (a).

[the 4th example of direct mode coding]

The 4th example of the coding processing of using direct mode then, is described.

The 4th example of the processing that Fig. 5 (b) expression is encoded to piece (object piece) BLa4 of picture (object picture) B11 in direct mode.

In the case, for object piece BLa4, based on as forward direction with reference to the selected immediate picture P10 of picture with as backward picture P13 with reference to picture, implement motion vector and be 0 bi-directional predicted processing.Motion vector MVf4, the MVh4 that uses when namely piece BLa4 being encoded is the expression picture motion vector with the zone relative same position of object piece BLa4 (piece) CRf4 P10, and the expression picture motion vector with the zone relative same position of object piece BLa4 (piece) CRh4 P13.

As mentioned above, in the direct mode shown in Fig. 5 (b) coding, the motion vector of object piece is set to 0 forcibly.Therefore, in the selecteed situation of direct mode, then there is no need to send the information of the motion vector of object piece, and do not need the calibration of motion vector to process, reduction that can the picked up signal treating capacity.The method also can be applied to, for example, as the backward of picture B11 with reference to the piece that is in piece BLa4 same position among the picture P13 of picture be, such as, do not had the such situation of piece of motion vector like that by the piece of intraframe coding.Thereby, even rear being in that piece with object piece same position does not have motion vector and in the situation of encoding, also can using direct mode and code efficiency is improved in the reference picture.

The processing of above-mentioned direct mode (the 1 4th example), not only can be applicable to picture displaying time be spaced apart fixing situation, also can be applicable to picture displaying time be spaced apart variable situation.

[the 5th example of direct mode coding]

Then, the 5th example of processing as the coding that uses direct mode being processed in the predictive coding under the direct mode of the situation of the variable spaced of the displaying time of picture describes.

Fig. 6 (a) is the figure that the 5th example that the coding of direct mode processes is used in explanation, and expression is processed the predictive coding under the direct mode as the 2nd example and is applied to processing in the situation of demonstration variable spaced of picture.

In the case, bi-directional predictive coding for the object piece BLa5 of object picture B11 is processed, utilization be in picture B11 the rear conduct with reference among picture (reference pictures) P13 of picture, be in motion vector (benchmark motion vector) MVf5 with piece (object piece) BLb5 of object piece BLa5 same position, process (the 2nd example) with the predictive coding under the direct mode shown in Fig. 4 (b) and similarly carry out.Motion vector MVf5 is the employed motion vector when encoding of the piece BLb5 to picture P13, the regional CRf5 of the corresponding blocks BLb5 of expression picture P7.In addition, parallel with motion vector MVf5 corresponding to motion vector MVg5, the MVh5 of object piece.These motion vectors MVg5, MVh5 are respectively the backward motion vector of regional CRh5 of the corresponding blocks BLa5 of the forward motion vector of regional CRg5 of corresponding blocks BLa5 of expression picture P10 and expression picture P13 in addition.

Also have, in the case, big or small MVF, the MVB of motion vector MVg5, MVh5 is also the same with the processing (the 2nd example) under the direct mode, can be obtained by formula (1), (2) respectively.

[jumping over specific processing]

Then, during the coding under direct mode processed, specific piece is described as jumping over the situation that piece processes.

In the coding under direct mode is processed, be in 0 the situation at the differential data of corresponding objects piece, in coded prediction error unit 103, can not be generated corresponding to the coded data of this object piece, do not export the bit string of corresponding objects piece from bit string generation unit 104.Differential data is 0 piece like this, processes as jumping over piece.

Below, specifically describe specific piece as the situation of jumping over piece and processing.

Fig. 6 (b) expression is as the specific picture F of the part of moving image.

In this picture F, the value of the differential data (prediction error data) of corresponding blocks MB (r-1), MB (r) and MB (r+3) is non-zero value in the piece MB of adjacency (r-1)～MB (r+3), and the value that correspondence is positioned at the differential data (prediction error data) of piece MB (r+1) between piece MB (r) and the piece MB (r+3) and piece MB (r+2) is 0.

In the case, under direct mode, piece MB (r+1) and piece MB (r+2) process as jumping over piece, in the bit string Bs of corresponding moving image, do not comprise the bit string corresponding to piece MB (r+1) and piece MB (r+2).

Fig. 6 (c) be for explanation with piece MB (r+1) and piece MB (r+2) as the figure that jumps over the flow structure in the situation that piece processes, shown in it corresponding to the part of piece MB (r) and the piece MB (r+3) of bit string Bs.

Between bit string Bmb (r) and the bit string Bmb (r+3) corresponding to piece MB (r+3) corresponding to piece MB (r), allocation list be shown in have between these pieces that two conducts jump over the processed piece of piece jump over identifier Sf (Sk:2).In addition, between bit string Bmb (r-1) and the bit string Bmb (r) corresponding to piece MB (r) corresponding to piece MB (r-1), allocation list is shown in not to be had between these pieces to jump over identifier Sf (Sk:0) as what jump over the processed piece of piece.

Bit string Bmb (r) corresponding to piece MB (r) is made of the Hmb of title section and data portion Dmb, in data portion Dmb, comprises the view data that is encoded corresponding to this piece.In addition, in the Hmb of title section, comprise the expression macro block (mb) type, namely this piece is the mode sign Fm that is encoded with what kind of coded system; Be illustrated in the coding time institute's reference picture with reference to image information Rp; And the information Bmvf, the Bmvb that are illustrated in employed motion vector when encoding.This piece MB (R) processes by bi-directional predictive coding and encodes, and information Bmvf, the Bmvb of expression motion vector, is illustrated respectively in the value of employed forward motion vector, backward motion vector in the bi-directional predicted processing.In addition, corresponding to corresponding other the bit string of piece such as the bit string Bmb (r+3) of piece MB (r+3), also have with corresponding to the identical structure of the bit string Bmb (r) of piece MB (r).

As mentioned above, by under direct mode, be that 0 piece is processed as jumping over piece with differential data, namely in bit string, also comprise mode information and jump over information corresponding to this piece, the reduction of size of code just becomes possibility.

Whether piece is jumped over, and can detect from the identifier Sf that jumps near front of the bit string that is disposed at each piece.In addition, whether piece is jumped over, and also can be learnt by the index information of recording and narrating the piece in the bit string of corresponding each piece etc.

In addition, processing example (the 1st example) under the direct mode shown in Fig. 4 (a), processing example (the 2nd example) under the direct mode shown in Fig. 4 (b), in the processing example (the 3rd example) under the direct mode shown in Fig. 5 (a), also can not to be that 0 piece is all processed as jumping over piece with differential data, but with relative object picture be positioned in time near front picture as forward direction with reference to picture, and be that 0 motion vector carries out bi-directional predicted for the object piece with size, only under the differential data of object piece is 0 situation, this object piece is processed as jumping over piece.

In addition, for the selection of the coded system of object piece, generally be to make encoding error corresponding to the quantity of predetermined position for minimum.Be output to bit string generation unit 104 by mode selected cell 109 determined coded systems.In addition, based on from the resulting prediction data of reference picture, being output to calculus of differences unit 102 and add operation unit 106 by mode selected cell 109 determined coded systems.But, in picture, in the selecteed situation of coding, export without prediction data.In addition, in the situation that mode selected cell 109 is selected to encode in the picture, make input terminal Ta be connected to lead-out terminal Tb1 by switch 111 is controlled, make lead-out terminal Td be connected to input terminal Tc1 by switch 112 is controlled.In the situation of selecting the inter-picture prediction coding, by being controlled, switch 111 make input terminal Ta be connected to lead-out terminal Tb2, and control by degree switch 112 and make lead-out terminal Td be connected to input terminal Tc2.

Below, the action by the dynamic image encoding device 10 in the situation of mode selected cell 109 selection inter-picture prediction codings is described.

Calculus of differences unit 102 receives from the prediction data Pd of mode selected cell 109 inputs.Calculus of differences unit 102, calculating is exported this differential data corresponding to the view data of the piece of picture B11 and the differential data of prediction data as prediction error data PEd.Prediction error data PEd is imported into coded prediction error unit 103.Generation coded data Ed is processed by the prediction error data PEd that inputs is implemented the codings such as frequency translation and quantification in coded prediction error unit 103.The coded data Ed of 103 outputs is imported into bit string generation unit 104 and predicated error decoding unit 105 from the coded prediction error unit.

Bit string generation unit 104 to the coded data Ed that inputs, is implemented Variable Length Code etc., and then the Information generation bit string Bs by additional movement vector and coded system etc., and exports this bit string Bs.At this moment, be to carry out in the situation of forward direction with reference to coded system in coded system, information (with reference to the image information) Rp with reference to which detection forward motion vector of picture P7, B9, P10 also is affixed among the bit string Bs for identification.

Then, use describing with reference to the management method of picture memory with reference to the adding method of image information in the process that Fig. 3 is encoded to picture B11.

When the coding of picture B11 begins, in reference picture memory 117, store picture P4, P7, P10, P13, B9.Picture B11 carries out bi-directional predictive coding with picture P13 as the candidate screen of backward reference with the candidate screen that picture P7, B9, P10 are used as the forward direction reference.The picture B11 that has encoded is stored in the storage area (#2) of having stored picture P4, because picture P4 is in the coding of the later picture of picture B11 is processed, not as using with reference to picture.

In the process that this picture B11 is encoded, be used for identification for the forward motion vector of object piece which the method for information (with reference to image information) with reference to picture P7, B9, P10 as additional, use from time apart from object picture (picture B11) nearest begin to give in order the method for index with reference to candidate screen.With reference to candidate screen, be when the object picture is encoded, to can be used as the picture of selecting with reference to picture.

Specifically, exactly picture P10 is given with reference to picture index [0], picture B9 is given with reference to picture index [1], and picture P7 is given with reference to picture index [2].

Thereby, when carrying out the object picture encoded with reference in the situation of picture P10, with reference to picture index [0] as expression with reference to the information of candidate screen in the tight the place ahead of object picture recorded and narrated in the bit string corresponding to the object piece.Similarly, with reference in the situation of picture B9, with reference to picture index [1] as expression with reference to the information of the candidate screen before 2 in the object picture, recorded and narrated in the bit string for the object piece.With reference in the situation of picture P7, with reference to picture index [2] as expression with reference to the information of the candidate screen before 3 in the object picture recorded and narrated in the bit string corresponding to the object piece.

For giving of the code of reference picture [0], [1], [2], be less according to the value of index, code length carries out with regard to shorter such mode.

Generally, candidate screen is used as the probability with reference to picture, and nearer candidate screen is just higher in time apart from the object picture.Thereby, by giving by this way code, be the overall code amount with reference to each code of which candidate screen in the time of just can reducing the motion vector of detection of expression object piece.

Predicated error decoding unit 105 for the coded data corresponding to the object piece of inputting, is implemented the decoding of re-quantization and frequency inverse conversion etc. and is processed, and generates the also decoding differential data PDd of object output piece.Decoding differential data PDd is by add operation unit 106 and prediction data Pd addition, and the decoded data Dd of the resulting object piece of addition is stored in reference to picture memory 117 thus.

By with above-mentioned same processing, to the processing of encoding of the rest block of picture B11.After the processing of whole pieces of picture B11 finishes, then carry out the coding of picture B12 and process.

The coding of＜picture B12 is processed 〉

Then, the coding processing of picture B12 described.

Because picture B12 is the B picture, so to the inter-picture prediction coding of the object piece among the picture B12, be to be positioned in time the place ahead or rear 2 predictive codings between the coded picture bi-directional picture of carrying out with reference to relative object picture.

Below, just used the situation of the inter-picture prediction coding of two-way reference to describe as the coding processing of picture B12.Thereby, in the case, as the candidate screen of forward direction with reference to picture, 2 pictures (I or P picture) that use relative object picture to approach by the displaying time order, perhaps the object picture is pressed the immediate B picture of displaying time order relatively.In addition, with reference to picture, use relative object picture by the immediate I of displaying time order or P picture as backward.Thereby, in the case, for picture B12 with reference to candidate screen, be exactly picture P7, P10, B11 (forward direction picture) and picture P13 (backward picture).

When to preparation when other pictures are encoded, when being used as encoding with reference to the B picture of picture, coding-control unit 110 is controlled each switch and is made switch 113,114,115 become on-state.Because picture B12 is used as with reference to picture when other pictures are encoded, so coding-control unit 110 is controlled each switch and made switch 113,114,115 become on-state.Thereby, 101 that read from the reference picture memory, corresponding to the view data of the piece of picture B12, be imported into motion vector detecting unit 108, mode selected cell 109 and calculus of differences unit 102.

Motion vector detecting unit 108, to be stored in reference to picture P7, P10, B11 in the picture memory 117, candidate screen as the forward direction reference, to be stored in reference to the picture P13 in the picture memory 117 as backward with reference to picture, carry out corresponding to the forward motion vector of the object piece of picture B12 and the detection of backward motion vector.

Detected motion vector is output to mode selected cell 109.

Mode selected cell 109 uses by motion vector detecting unit 108 detected motion vectors, determines the coded system of the object piece of picture B12.In this occasion, as the coded system to B picture B12, in picture, encode, use the inter-picture prediction coding of forward motion vector, the inter-picture prediction coding of use backward motion vector, the inter-picture prediction coding of use bi-directional motion vector, and select in the direct mode.In the situation that coded system is encoded for the inter-picture prediction of using forward motion vector, from picture P7, P10, B11, select optimal picture as the reference picture.

Below, illustrate in direct mode the processing of encoding of the piece of picture B12.

[the 1st example of direct mode coding]

The situation that Fig. 7 (a) expression is encoded to piece (object piece) BLa5 of picture (object picture) B12 in direct mode.This direct mode coding utilize picture B12 the rear conduct with reference among picture (reference pictures) P13 of picture, be in motion vector (benchmark motion vector) MVc5 with piece (reference block) BLb5 of object piece BLa5 same position.Motion vector MVc5 is piece BLb5 employed motion vector when being encoded, and is stored in the motion vector storage unit 116.This motion vector MVc5, the regional CRc5 of the piece BLb5 that expression picture P10 is corresponding.To piece BLa5, use the motion vector parallel with motion vector MVc5, based on picture B11, the P13 with reference to picture as piece BLa5, implement bi-directional predictive coding and process.The motion vector that uses when piece BLa5 is encoded is the motion vector MVd5 of the regional CRd5 of the corresponding blocks BLa5 among the expression picture B11, and the motion vector MVe5 of the regional CRe5 of the corresponding blocks BLa5 among the expression picture P13.Big or small MVF, the MVB of motion vector MVd5, MVe5 can be obtained by above-mentioned formula (1), (2) respectively.

[the 2nd example of direct mode coding]

The 2nd example of the coding processing of using direct mode then, is described.

The situation that Fig. 7 (b) expression is encoded to piece (object piece) BLa6 of picture (object picture) B12 in direct mode.The conduct that this direct mode coding utilizes the rear that is in picture B12 with reference among picture (reference pictures) P13 of picture, be in motion vector (benchmark motion vector) MVf6 with piece (reference block) BLb6 of object piece BLa6 same position.Motion vector MVc6 is piece BLb6 employed motion vector when being encoded, and is stored in the motion vector storage unit 116.This motion vector MVf6, expression picture P7 is corresponding to the regional CRf6 of piece BLb6.To piece BLa6, use the motion vector parallel with motion vector MVc6, based on as picture B11 and picture P13 with reference to picture, implement bi-directional predictive coding and process.The motion vector that uses when piece BLa6 is encoded is that expression is the motion vector MVg6 that represents the regional CRg6 of the corresponding blocks BLa6 of picture B11, and the motion vector MVh6 of the regional CRh6 of the corresponding blocks BLa6 of expression picture P13.Big or small MVF and the MVB of motion vector MVg6 and MVh6 can be obtained by above-mentioned formula (1), (2) respectively.

As mentioned above, in direct mode, to when object piece BLa6 is encoded as rear picture to reference, be positioned at this object piece relatively the motion vector MVf6 of the BLb6 of identical position calibrate, thereby obtain forward motion vector MVg6 and backward motion vector MVh6 for the object piece.So, in the selecteed situation of direct mode, just do not need to send the information of the motion vector of object piece.And, since as forward direction with reference to picture, use and press the immediate coded picture of displaying time ordinal position apart from the object picture, so can make the forecasting efficiency raising.

[the 3rd example of direct mode coding]

The 3rd example of the coding processing of using direct mode then, is described.

The 3rd example that Fig. 8 (a) expression is encoded and processed piece (object piece) BLa7 among picture (object picture) B12 in direct mode.

This direct mode coding utilize the backward conduct that is in picture B12 with reference among picture (reference pictures) P13 of picture, be in motion vector (benchmark motion vector) MVc7 with piece (reference block) BLb7 of piece BLa7 same position.Motion vector MVc7 is employed motion vector when piece BLb7 is encoded, and is stored in the motion vector storage unit 116.This motion vector MVc7, the regional CRc7 of the corresponding blocks BLb7 among the expression picture P7.To piece BLa7, use the motion vector parallel with motion vector MVc7, will with piece BLb7 be encoded the identical picture (picture P7) of the picture of time institute's reference as forward direction with reference to picture, with picture P13 as backward with reference to picture, implement bi-directional predictive coding and process.The motion vector that uses when piece BLa7 is encoded is the motion vector MVe7 of the regional CRe7 of the corresponding blocks BLa7 of the motion vector MVd7 of the regional CRd7 of the corresponding blocks BLa7 of expression picture P7 and expression picture P13.

Big or small MVF, the MVB of above-mentioned motion vector MVd7 and MVe7 can be obtained by above-mentioned formula (2), (3) respectively.

When the picture of the time institute's reference that is encoded at piece BLb7 in the situation of reference picture memory 117 deletions, immediate forward direction is used as forward direction in the direct mode with reference to picture with reference to picture in time.Direct mode coding in the case is identical with the 1st example of direct mode coding.

As mentioned above, in the coding of the direct mode shown in Fig. 8 (a), to when the object piece is encoded as rear picture to reference, be positioned at this object piece relatively the motion vector MVf7 of the piece BLb7 of identical position calibrate, thereby obtain forward motion vector MVd7 and backward motion vector MVe7 corresponding to the object piece.So, in the selecteed situation of direct mode, just do not need to send the information of the motion vector of object piece.

[the 4th example of direct mode coding]

The 4th example of the coding processing of using direct mode then, is described.

The 4th example of the processing that Fig. 8 (b) expression is encoded to piece (object piece) BLa8 of picture (object picture) B12 in direct mode.

In the case, for object piece BLa8, based on as forward direction with reference to the selected immediate picture P10 of picture with as backward picture P13 with reference to picture, implement motion vector and be 0 bi-directional predicted processing.That is motion vector MVf8, the MVh8 that uses when, piece BLa8 being encoded be respectively among the expression picture P11 with object piece BLa8 relatively among the motion vector of the zone of same position (piece) CRf8 and the expression picture P13 with the object piece BLa8 motion vector of the zone of same position (piece) CRh8 relatively.

As mentioned above, in the direct mode shown in Fig. 8 (b) coding, the motion vector of object piece is set to 0 forcibly.Therefore, in the selecteed situation of direct mode, then there is no need to send the information of the motion vector of object piece, and do not need the calibration of motion vector to process, reduction that can the picked up signal treating capacity.The method also can be applied to, for example, as the backward of picture B12 with reference to the piece that is in piece BLa8 same position among the picture P13 of picture be, such as, do not had the such situation of piece of motion vector like that by the piece of intraframe coding.Thereby, even in the situation that does not have motion vector ground to encode at the rear piece that is in the reference picture with object piece same position, also can use direct mode and code efficiency is improved.

Processing under the direct mode of above-mentioned picture B12 (the 1st～the 4th example), not only can be applicable to picture DISPLAY ORDER be spaced apart fixing situation, also can be applicable to picture DISPLAY ORDER be spaced apart variable situation, the situation of the picture B11 shown in Fig. 6 (a).

In addition, during coding under the direct mode of picture B12 is processed, process same with the coding under the direct mode of picture B11, be in 0 the situation at the differential data corresponding to the object piece, coded prediction error unit 103 does not generate the coded data corresponding to this object piece, does not export bit string corresponding to the object piece from bit string generation unit 104.Like this, differential data is 0 piece, and is the same with the situation of the picture B11 shown in Fig. 6 (b), Fig. 6 (c), processes as jumping over piece.

In addition, in processing example (the 2nd example) under the direct mode shown in processing under the direct mode shown in Fig. 7 (a) example (the 1st example), Fig. 7 (b), the processing example (the 3rd example) under the direct mode shown in Fig. 8 (a), also can not to be that 0 piece is all processed as jumping over piece with differential data.In other words, relative object picture to be positioned in time near front picture be used as forward direction with reference to picture, and be that 0 motion vector carries out bi-directional predicted for the object piece with size, and only under the differential data of object piece is 0 situation, this object piece is processed as jumping over piece.

When the mode selected cell 109 of coded system determined by to(for) the object piece of picture B12, process the same with the coding of the object piece of picture B11, for object piece generation forecast data PEd, this prediction data PEd is output to calculus of differences unit 102 and add operation unit 106.But, in the selecteed situation of in picture, encoding, from mode selected cell 109 not prediction of output data.In addition, be to carry out as coded system according to which that selected by mode selected cell 109 to encode between coding and picture in the pictures to each switch 111,112 control, the same with the control of the coding situation of picture B11.

Below, when being encoded, picture P12 selected the action of the dynamic image encoding device 10 in the situation of inter-picture prediction coding to describe by mode selected cell 109.

In the case, calculus of differences unit 102, coded prediction error unit 103, bit string generation unit 104, predicated error decoding unit 105, add operation unit 106 and with reference to picture memory 117 and have selected the situation of inter-picture prediction coding to move by mode selected cell 109 when picture P11 is encoded the samely.

But, in the case, because the candidate screen of forward direction reference is different from the situation of the coding of picture P11, so be when carrying out the coded system of forward direction reference in the coded system for the object piece, be additional to the object piece bit string to become for the identification forward motion vector with reference to image information be which information with reference to picture P7, P10, B11.

In addition, when picture B12 is encoded, with reference to the picture memory management method with reference to the adding method of image information, just the same with method when picture B11 is encoded shown in Figure 3.

As mentioned above, in embodiments of the present invention 1, B picture (object picture) when encoding, as the candidate screen of forward direction reference, is also being used the B picture except the P picture.So, as for the object B picture with reference to picture, use the immediate forward direction picture be positioned at this B picture just to become possibility, thereby can improve the precision of prediction of the motion compensation of B picture, can obtain the raising of code efficiency.

In execution mode 1, in the process that the P picture is encoded, do not use the B picture as the reference picture.Therefore, even in the middle of decoding, produced in the wrong situation, also can restart decoding by next I or the P picture from the picture that produced decoding error, fully carry out the recovery from mistake.But, even in the process that the P picture is encoded with the B picture as with reference to picture, can not changed by the effect of obtained other of present embodiment yet.

In addition, because the candidate screen with the forward direction reference in the B picture codec is made as 2 P pictures and 1 B picture, so when with the existing situation that will be made as for the candidate of the forward direction reference of B picture 3 P pictures relatively the time, do not change in the quantity for the candidate screen of the forward direction reference of B picture.Therefore, can avoid, storage that the B picture causes owing in the candidate screen for the forward direction reference of B picture, comprising to increase with reference to the capacity with reference to picture memory of candidate screen and the increase of the treating capacity of motion detection.

In addition, in present embodiment 1, expression is implemented the information that the inter-picture prediction coding is processed to the B picture with reference to forward direction B picture, and be illustrated in this forward direction with reference to the time be with reference to the information of several I or P picture and several B pictures, record and narrate the heading message for the bit string that will generate.Therefore, just can understand when the bit string that is generated by dynamic image encoding device is decoded, need which kind of capacity with reference to picture memory.

In addition, in bit string during the information such as additional movement vector and coded system, to carry out in the situation of coded system of forward direction reference in coded system, in bit string additional to give as the candidate screen of forward direction reference, be used for identification with reference to picture with reference to image information, and and then the management method with reference to picture memory of using according to the reference picture, to distance object picture in time nearest candidate screen give with reference to image information, represent with short code length.So, just can reduce to express the size of code with reference to the integral body of the code of image information.In addition, in the management of reference picture memory, for example because it doesn't matter that ground manages the reference picture memory with the kind of picture, so can make with reference to the picture memory amount minimum.

In addition, in this execution mode 1, carry out with reference to the management of picture with reference to picture memory in the zone by separating the P picture and the zone of B picture, just becomes simple and easy with reference to the management of picture memory.

In addition, when in direct mode the piece of B picture being encoded, using relatively as forward direction with reference to picture, this B picture just can make thus for the forecasting efficiency of B picture under direct mode and improve by the immediate picture of time DISPLAY ORDER.

In addition, when in direct mode the piece of B picture being encoded, as forward direction with reference to picture use rear when being encoded to the reference picture by the picture of forward direction reference, just can make thus for the forecasting efficiency raising of B picture under direct mode.

In addition, when in direct mode the piece of B picture being encoded, based on forward direction with reference to picture and backward with reference to picture carry out with motion vector be made as 0 bi-directional predicted, so just do not need the calibration of the motion vector under the direct mode to process, reduction that can the acquired information treating capacity.At this moment, even under being in piece with rear object piece same position in the reference picture and not having the situation of motion vector, encode, also can use direct mode that code efficiency is improved.

In addition, when in direct mode the piece of B picture being encoded, if be 0 for the predicated error of object piece, in bit string, do not record and narrate the information about this object piece, so the reduction of size of code just becomes possibility.

In this execution mode 1, motion compensation be take each by image space (macro block) that 16 pixels of level * vertical 16 pixels form as unit, and the coding of prediction error image is to be processed as unit by image space (sub-block) that 8 pixels of level * vertical 8 pixels form take each.But the pixel count of each macro block (sub-block) in the motion compensation (coding of prediction error image) also can be from shown in the above-mentioned execution mode 1 different.

In addition, although in this execution mode 1, be illustrated take the number of continuous B picture as 2 situation as example, the number of continuous B picture also can be more than or equal to 3.

For example, between I picture and P picture, perhaps the number of the B picture between 2 P pictures also can be 3 or 4.

In addition, in this execution mode 1, coded system to the P picture is to encode, used the inter-picture prediction coding of motion vector and do not use the inter-picture prediction coding of motion vector to select in picture, and the coded system of B picture be in the picture coding, used forward motion vector the inter-picture prediction coding, used backward motion vector the inter-picture prediction coding, used the inter-picture prediction coding of bi-directional motion vector, and select among the direct mode.But the coded system of P picture or B picture also can be the mode beyond shown in the above-mentioned execution mode 1.

For example, do not using in the situation of direct mode as the coded system of B picture, just do not needing the motion vector storage unit 116 in the dynamic image encoding device 10.

In addition, although in this execution mode 1, be the candidate screen with reference to picture when other pictures are encoded as the picture B11 (B12) of B picture, but when other pictures are encoded,, just there is no need to be stored in reference in the picture memory 117 during as the reference picture without the B picture.In the case, coding-control unit 110 disconnects switch 114,115.

In addition, although in this execution mode 1, the quantity of the candidate screen of the forward direction reference of using in the process that the P picture is encoded is 3, is not limited to this.For example, in the process that the P picture is encoded, also can use 2 or more than or equal to 4 the picture candidate screen as the forward direction reference.

In addition, although in this execution mode 1, the candidate screen of the forward direction reference of using in the process that the B picture is encoded is 2 P pictures and 1 B picture, but the candidate screen of the forward direction reference in the process that the B picture is encoded is not limited to the situation of above-mentioned execution mode 1.

For example, in the process that the B picture is encoded, also can use 1 P picture and 2 B pictures, 2 P pictures and 2 B pictures, and no matter in time immediate 3 pictures of picture type as the candidate screen of forward direction reference.In addition, also can not with immediate B picture on the displaying time axle, but will use as the candidate screen of reference at the B picture that the displaying time axle leaves the object picture.

In addition, when when the piece of B picture is encoded, with reference to 1 backward picture, only in the situation of a picture as the candidate screen of forward direction reference near the object picture, the indicated object piece is just to there is no need to record and narrate in bit string with reference to the information (with reference to image information) which picture is encoded.

In addition, in this execution mode 1, when the B picture is encoded, with reference to the B picture in more the place ahead of the immediate forward direction P picture that is located at the object picture.But when the B picture is encoded, needn't be with reference to the B picture in more the place ahead of the immediate forward direction I that is located at the object picture or P picture.At this moment, when the bit string that generates is decoded, even in the middle of decoding, produce mistake, also can restart decoding by next I or the P picture from the picture that produces decoding error, can fully carry out the recovery from mistake.

For example, Fig. 9 (a) and Fig. 9 (b) are when being illustrated in the B picture and being encoded, with reference to the figure of the situation of the B picture in more the place ahead of the immediate forward direction P picture that is located at the object picture.

Fig. 9 (a) expression picture is arranged and B picture and with reference to an example of the relation of picture.Particularly, the number of B picture of expression between the P of adjacency picture is 2 in Fig. 9 (a), is the situation of 1 P picture and 2 B pictures for the forward direction of B picture with reference to the candidate screen of picture (namely B picture be encoded the picture of time institute's reference).

Fig. 9 (b) expression picture is arranged and B picture and with reference to another example of the relation of picture.Particularly, the number of B picture of expression between the P picture is 4 in Fig. 9 (b), for the forward direction of B picture with reference to the candidate screen of picture for the situation of picture type it doesn't matter ground is positioned in time immediate 2 pictures of object picture.

When Figure 10 (a) and Figure 10 (b) are illustrated in the B picture and are encoded, not with reference to the situation of the B picture in more the place ahead of immediate forward direction I or P picture.

Particularly, in Figure 10 (a), the number of the B picture of expression between the P picture is 2, be 1 P picture and 1 B picture for the forward direction of B picture with reference to the candidate screen of picture, be located at the B picture in more the place ahead of immediate forward direction P picture of object picture not as the situation of above-mentioned forward direction with reference to the candidate screen of picture.

In Figure 10 (b), the number of the B picture of expression between the P picture is 4, be 1 P picture and 1 B picture for the forward direction of B picture with reference to the candidate screen of picture, be located at the B picture in more the place ahead of immediate forward direction P picture of object picture not as the situation of above-mentioned forward direction with reference to the candidate screen of picture.

In addition, in this execution mode 1, the quantity with reference to candidate screen of using for the P picture is 3, the candidate screen of the forward direction reference of using for the B picture is 2 P pictures and 1 B picture, but frame numbers that namely can reference when the P picture is encoded is identical with the frame numbers of forward direction reference when the B picture is encoded, but but the frame numbers that the frame numbers of forward direction reference also can reference than when the P picture is encoded when the B picture is encoded lack.

In addition, in this execution mode 1, the example as the coding of direct mode is processed is illustrated 4 methods, but can use in these methods one in direct mode, also can use a plurality of in 4 methods.But in the situation of using a plurality of methods, preferably will represent to use the information (DM mode information) of which direct mode to record and narrate in bit string.

For example, in the situation that spreads all over a method of the whole use of sequence, DM mode information is recorded and narrated in the title of sequence integral body.In the situation to a method of each picture and selecting, DM mode information is recorded and narrated in the title of this picture.In the situation to a method of each piece selection, DM mode information is recorded and narrated in the title of piece.

Although as the unit of selecting the coding method under the above-mentioned direct mode, picture or piece, but also can be the GOP (Group of pictures) that comprises a number of frames, by the picture bar of a plurality of GOB that form (Group of blocks), divided frame etc.

In addition, although in this execution mode 1, the management method as the reference picture memory describes with reference to Fig. 3, is not limited to shown in Figure 3 with reference to the management method of picture memory.

Below, other the management method with reference to picture memory is described.

At first, to becoming the 1st example in the situation that P picture and B picture manage as the employed whole picture segmentation of reference picture, describe with reference to Figure 11.

In the case, establish with reference to picture memory 117, as storage area, have the zone of P picture storage area (#1)～(#4), B picture storage area (#1), 6 picture amounts of (#2) total.The memory cell of each picture is not limited to one with reference to the zone in the picture memory, also can be a memory.

When the coding of picture P13 begins, in the P of reference picture memory 117 picture storage area (#1)～(#4), store respectively picture P1, P4, P7, P10, and in this B picture storage area (#1), (#2), store respectively picture B8, B9.Picture P13 is used as picture P4, P7, P10 to encode with reference to the candidate screen of picture, and the picture P13 that storage has been encoded in having stored the zone (#1) of picture P1, because picture P1 is used as with reference to picture at picture P13 and later picture thereof the Shi Buhui that is encoded.

In the case, give picture P4, the P7 as candidate screen, the method for P10 with reference to image information, with method shown in Figure 3 similarly to relative object picture in time nearer candidate screen value of giving less with reference to picture index.

Particularly, the hithermost forward direction candidate screen of distance object picture is given with reference to picture index [0], distance object picture the 2nd close candidate screen is given with reference to picture index [1], and distance object picture candidate screen is farthest given with reference to picture index [2].

In Figure 11, to being used as backward picture with reference to picture, give symbol [b] as the reference image information, when object picture and later picture thereof are encoded, to being not used as the picture with reference to picture, give symbol [n].

Then, to being divided into as whole pictures that the reference picture uses the 2nd example in the situation that P picture and B picture manage, describe with reference to Figure 12.

Because the storage management in this 2nd example is identical with the 1st example shown in Figure 11, so the description thereof will be omitted.

In this 2nd example, as the adding method of reference picture index, make the picture that is stored in the P picture storage area preferentially carry out the method that index is given.But because in the process that picture P13 is encoded, the B picture is not used as with reference to picture, gives so the B picture is not carried out index.Thereby, picture P10 is given with reference to picture index [0], picture P7 is given with reference to picture index [1], picture P4 is given with reference to picture index [2].

Then, to being divided into the 3rd example in the situation that P picture and B picture manage as the employed whole pictures of reference picture, describe with reference to Figure 13.

Because the storage management in this 3rd example is identical with the 1st example shown in Figure 11, so the description thereof will be omitted.

In this 3rd example, as the adding method of reference picture index, make the picture that is stored in the B picture storage area preferentially carry out the method that index is given.But because in the process that picture P13 is encoded, the B picture is not used as with reference to picture, gives so the B picture is not carried out index.Thereby, picture P10 is given with reference to picture index [0], picture P7 is given with reference to picture index [1], picture P4 is given with reference to picture index [2].

Then, to being divided into the 4th example in the situation that P picture and B picture manage as the employed whole pictures of reference picture, describe with reference to Figure 14.

Because the storage management in this 4th example is identical with the 1st example shown in Figure 11, so the description thereof will be omitted.

In this 4th example, as the adding method of reference image information, use each picture to encoding, determine it is to make the picture of among which of P picture storage area and B picture storage area, storing preferential, give the method with reference to picture index.

Specifically, for example, according to distance object picture hithermost kind with reference to picture in time, determine that make the picture of storing preferentially carry out index gives in which storage area of P picture storage area and B picture storage area.

In the process that picture P13 is encoded, because the B picture is not used as with reference to picture, give so make the picture of in P picture storage area, storing preferentially carry out index.Thereby, picture P10 is given with reference to picture index [0], picture P7 is given with reference to picture index [1], picture P4 is given with reference to picture index [2].In the case, make the picture of P picture storage area preferentially carry out the information of giving with reference to picture index expression, record and narrate in the title of picture.

In the adding method with reference to picture index of above-mentioned Fig. 3 and Figure 11～shown in Figure 14, less with reference to the value of picture index, expression is just shorter with reference to the code length of the code of picture index.General since relatively the object picture in time the closer to picture be used as with reference to the probability of picture higher, so by determining like this expression with reference to the code length of the code of picture index, just can reduce to express the quantity with reference to the code of the integral body of the code of picture index.

Although about the management method of reference picture memory with reference to the adding method of picture index, show Fig. 3, Figure 11～5 methods shown in Figure 14, also any one in these 5 methods of choice for use in advance.Also can switch a plurality of methods in these 5 methods of use.Which but switching in the situation of using a plurality of methods, preferably will record and narrate as heading message etc. with the information of method and so on.

In addition, by as heading message, record and narrate expression P picture and carry out the information that inter-picture prediction is encoded with 3 with reference to candidate screen, just can understand when the bit string Bs that the dynamic image encoding device 10 by present embodiment 1 is generated decodes, need which kind of capacity with reference to picture memory.These heading messages also can be recorded and narrated in the title of sequence integral body, also can record and narrate in the title of each GOP that comprises a plurality of pictures (Group of pictures), perhaps in the title of each picture.

Below, that uses when being encoded as picture B11 describes with reference to the management method of picture memory with reference to the method beyond the method shown in Figure 3 of the adding method of image information (being divided into the method that P picture and B picture manage with reference to candidate screen).

At first, use Figure 11, the 1st example that the B picture that is divided into the P picture with reference to candidate screen is managed is explained.

When the coding of picture B11 began, in reference picture memory 117, picture P4, P7, P10, P13 were stored in P picture storage area, and B8, B9 are stored in B picture storage area.Then picture B11 encodes with picture P13 the candidate screen that picture P7, B9, P10 are used as the forward direction reference as the candidate screen of backward reference.The picture B11 that has encoded is stored in the storage area of having stored picture P8, because picture B8 is in the process that picture B11 and later picture thereof are encoded, as not using with reference to picture.

In the case, as the method for giving each picture with reference to image information (namely be used for being illustrated in when detecting forward motion vector is to use with reference to which of picture P7, B9, P10 with reference to the information of candidate screen), similarly use the method for in order the reference candidate screen being given index apart from the nearest beginning of object picture from time with the example of adding method shown in Figure 3.

Namely, candidate screen (picture P10) to the tight the place ahead of object picture (picture B11) is given with reference to picture index [0], candidate screen (picture B9) before 2 in the object picture is given with reference to picture index [1], and the candidate screen (picture P7) before 3 in the object picture is given with reference to picture index [2].

Then, use Figure 12 to when picture B11 is encoded, be divided into the 2nd example that P picture and B picture manage with reference to candidate screen and be explained.

Because in this 2nd example, storage management is identical with the 1st example illustrated in fig. 11, so the description thereof will be omitted.

In this 2nd example, as the adding method of reference picture index, make the picture that is stored in the P picture storage area preferentially carry out the method that index is given.Thereby, picture P10 is given with reference to picture index [0], picture P7 is given with reference to picture index [1], picture B9 is given with reference to picture index [2].

Then, use Figure 13 to when picture B11 is encoded, be divided into the 3rd example that P picture and B picture manage with reference to candidate screen and be explained.

Because in this 3rd example, storage management is identical with the 1st example illustrated in fig. 11, so the description thereof will be omitted.

In this 3rd example, as the adding method of reference picture index, make the picture that is stored in the B picture storage area preferentially carry out the method that index is given.Thereby, picture B9 is given with reference to picture index [0], picture P10 is given with reference to picture index [1], picture P7 is given with reference to picture index [2].

Then, use Figure 14 to when picture B11 is encoded, be divided into the 4th example that P picture and B picture manage with reference to candidate screen and be explained.

Because in this 4th example, storage management is identical with the 1st example illustrated in fig. 11, so the description thereof will be omitted.

In this 4th example, as the adding method of reference picture index, use each picture of encoding is determined it is to make the picture of storing preferential in which zone of P picture storage area and B picture storage area, give the method with reference to image information.

Specifically, be according to the object picture that relatively becomes coded object hithermost kind with reference to candidate screen in time, determine which memory is preferentially carried out index gives.

In the process that picture B11 is encoded and since relatively the object picture in time hithermost forward direction be picture P10 with reference to picture, give so the picture of storing in P picture storage area is preferentially carried out index.

Thereby, picture P10 is given with reference to picture index [0], picture P7 is given with reference to picture index [1], picture B9 is given with reference to picture index [2].In the case, make the picture of P picture storage area preferentially carry out the information of giving with reference to picture index expression, record and narrate in the title of picture.

In the method for when picture B11 is encoded, the reference picture index being given (Fig. 3, Figure 11～5 methods shown in Figure 14), the same with the situation of the coding of picture P13, value with reference to picture index is less, and expression is just shorter with reference to the code length of the code of picture index.

In addition, the same with the situation of the coding of P picture P13 in the process that B picture B11 is encoded, also any one in above-mentioned 5 methods of choice for use in advance.Also can switch a plurality of methods in these 5 methods of use.But switch to use in the situation of a plurality of methods, preferably should record and narrate as heading message with information of which method and so on etc.

In addition, by as heading message, recording and narrating expression B picture processes forward direction B picture as implement the inter-picture prediction coding with reference to candidate screen, and the candidate screen of the forward direction reference of using when the B picture is encoded is, the information of 2 I or P picture and 1 B picture, just can understand when the bit string that the dynamic image encoding device 10 by present embodiment 1 is generated is decoded, need to have which kind of memory capacity with reference to picture memory.These heading messages also can be recorded and narrated in the title of sequence integral body, also can record and narrate in the title of each GOP that comprises a plurality of pictures (Group of pictures), perhaps in the title of each picture.

That uses when being encoded as picture B12 at last, describes with reference to the management method of picture memory with reference to the method beyond the method shown in Figure 3 of the adding method of image information (being divided into the method that P picture and B picture manage with reference to candidate screen).

Because Figure 11～it is shown in Figure 13 that the 1st example～the 3rd example is identical with the situation of the coding of picture B11, so the description thereof will be omitted.

Therefore, about picture B12, use Figure 14 only to illustrate with reference to candidate screen and be divided into the 4th example that P picture and B picture manage.

In this 4th example, about storage management, because with to be divided into the 1st example that P picture and B picture manage with reference to candidate screen when picture B11 is encoded identical, so the description thereof will be omitted.

In this 4th example, to detect forward motion vector be the method for giving each picture with reference to which the information of candidate screen P7, P10 and B11 about being illustrated in, should use each picture to encoding, determine it is to make the preferential method of candidate screen of in which zone of P picture storage area and B picture storage area, storing.

Specifically for example, in the process that picture B12 is encoded, according to distance object picture hithermost kind with reference to picture in time, determine to make the candidate screen of P picture storage area and which storage area of B picture storage area preferentially to carry out cross index and give.

In the process that picture B12 is encoded, owing to the candidate screen apart from time hithermost forward direction reference of object picture (picture B12) is picture B11, so preferentially being carried out index, gives in the picture of storing in B picture storage area.

Thereby, picture B11 is given with reference to picture index [0], picture P10 is given with reference to picture index [1], picture P7 is given with reference to picture index [2].In the case, make the picture of B picture storage area preferentially carry out the information of giving with reference to picture index expression, record and narrate in the title of each picture.

In addition, as heading message, the same with the situation that picture B11 is encoded, record and narrate expression B picture and also forward direction B picture is processed as implement the inter-picture prediction coding with reference to candidate screen, and the candidate screen of the forward direction reference of using when the B picture is encoded is the information of 2 I or P picture and 1 B picture.

In addition, in this execution mode 1, just the quantity with reference to candidate screen for the P picture is 3, is the situation of 2 P pictures and 1 B picture for the candidate screen of the forward direction reference of B picture, and 5 examples with reference to the management method of picture memory (Fig. 3, Figure 11～Figure 14) have been described.But also can be applied to respectively with reference to the different situation of the quantity of candidate screen and execution mode 1 with reference to 5 examples of the management method of picture memory.In the different situation of the quantity of reference candidate screen, different from execution mode 1 with reference to the capacity of picture memory.

In addition, in this execution mode 1, being distinguished into reference to candidate screen in the method (4 examples shown in Figure 11～Figure 14) that P picture and B picture manage the reference picture memory, store the P picture into P picture storage area, store the B picture into B picture storage area.But in the memory cell of stored picture, also can utilize defined short period picture memory and long period picture memory in H.263++.For example, have respectively the short period picture memory is utilized as P picture storage area, the method that the long period picture memory is utilized as B picture storage area.

[execution mode 2]

The below describes embodiments of the present invention 2.

Figure 15 is the block diagram that explanation utilizes the moving image decoding apparatus of embodiments of the present invention 2 to use.

Moving image decoding apparatus 20 is decoded to the bit string Bs that exports from the dynamic image encoding device 10 of execution mode 1.

Particularly, this moving image decoding apparatus 20 has the bit string analytic unit 201 that the above-mentioned bit string Bs of analysis carries out the extraction of various data; Coded data Ed from 201 outputs of this bit string analytic unit is decoded and the predicated error decoding unit 202 of prediction of output error information PDd; And based on selecting the mode decoding unit 223 of relevant mode information (coded system) Ms output switch control signal Cs with the mode that is extracted by this bit string analytic unit 201.

Moving image decoding apparatus 20 also has the view data DId that maintenance is decoded, with the view data of storing as comparable data Rd or output image data Od export with reference to picture memory 207; Based on the information of the motion vector MV that extracts from these data of reading with reference to picture memory 207 (reference image data) Rd, by above-mentioned bit string analytic unit 201 and from the coded system Ms of aforesaid way decoding unit 223, the motion compensation decoding unit 205 of generation forecast data Pd; And above-mentioned prediction data Pd is added to the output data PDd of above-mentioned predicated error decoding unit 202, the add operation unit 208 of generating solution code data Ad.

Moving image decoding apparatus 20 also has based on by 201 extracting header information of above-mentioned bit string analytic unit Ih, controls memory control unit 204 with reference to picture memory 207 by memory control signal Cm; Be arranged on the selector switch 209 between above-mentioned predicated error decoding unit 202 and the add operation unit 208; And be arranged on this add operation unit 208 and with reference to the selector switch 210 between the picture memory 207.

Above-mentioned selector switch 209 has an input terminal Te and two lead-out terminal Tf1 and Tf2, and according to above-mentioned switch controlling signal Cs, this input terminal Te is connected to the side of above-mentioned two lead-out terminal Tf1, Tf2.Selector switch 210 has two input terminal Tg1 and Tg2 and a lead-out terminal Th, and according to above-mentioned switch controlling signal Cs, this lead-out terminal Th is connected to the side of above-mentioned two input terminal Tg1, Tg2.In addition, by above-mentioned selector switch 209, input terminal Te is applied the output data PDd of predicated error decoding unit 202, from a side lead-out terminal Tf1 the input terminal Tg1 of selector switch 210 is exported the output data PDd of above-mentioned predicated error decoding unit 202, from the opposing party's the lead-out terminal Tf2 output data PDd to the above-mentioned predicated error decoding unit 202 of add operation unit 208 outputs.By above-mentioned selector switch 210, one side's input terminal Tg1 is inputted the output data PDd of above-mentioned predicated error decoding unit 202, the opposing party's input terminal Tg2 is inputted the output data Ad of add operation unit 208, from lead-out terminal Th, the one party of this output data PDd and Ad is output to reference to picture memory 207 as decode image data DId.

In addition, moving image decoding apparatus 20 has the motion vector MV that keeps from motion compensation decoding unit 205, the motion vector storage unit 226 that the motion vector MV that stores is exported to above-mentioned motion compensation decoding unit 205.

The below describes the action of moving image decoding apparatus 20.

In the following description, object picture relatively to be decoded is positioned at the picture at the place ahead or rear at the displaying time axle, refers to that relative object picture is in the picture at the place ahead or rear in time, perhaps also is called simply forward direction or backward picture.

In moving image decoding apparatus shown in Figure 15 20, the bit string Bs that input is generated by the dynamic image encoding device 10 of execution mode 1.In this execution mode 2, the bit string of P picture is positioned in time selected 1 of 3 candidate screen (I or P picture) nearby at the place ahead or rear with reference to relative P picture and implements the inter-picture prediction coding and obtain.In addition, the bit string of B picture with reference to relative this B picture be arranged in the place ahead or rear 4 candidate screen (namely be in time 2 I in immediate the place ahead or P picture, in time be in immediate 1 B picture and be positioned in time I or the P picture at rear) selected 2 implement the inter-picture prediction coding and obtain.In for 4 candidate screen as the B picture of processing object, comprise another B picture that relative this object B picture is in the place ahead in time.

In addition, P picture or B picture are with reference to which candidate screen to encode, and the heading message that can be used as bit string is recorded and narrated.Thereby, be with reference to just can learning by extracting this heading message by bit string analytic unit 201 which picture carries out about the coding of processing the object picture.This heading message Ih is also to memory control unit 204 outputs.

In the case, corresponding with the picture in bit string coded data shown in Figure 16 (a), is arranged according to coded sequence.

Specifically, the coded data of the picture among the bit string Bs is arranged by the order of picture P4, B2, B3, P7, B5, B6, P10, B8, B9, P13, B11, B12, P16, B14, B15.In other words, the arrangement of this picture becomes from the upper decode time of the decode time axle Y beginning early of expression decode time Tdec and in order each picture has been carried out (pressing decoding order arranges) arranged sequentially.

Figure 16 (b), the picture that expression will be arranged by above-mentioned decoding order changes the arrangement of lining up by the picture of DISPLAY ORDER.Namely, in Figure 16 (b), the arrangement of picture B2, B3, P4, B5, B6, P7, B8, B9, P10, B11, B12, P13, B14, B15, P16 is in order each picture to be carried out (pressing DISPLAY ORDER arranges) arranged sequentially from the upper displaying time order of the displaying time axle X beginning early of expression displaying time Tdis.

Below, the decoding of picture P13, B11, B12 processed describing in this order.

The decoding of＜picture P13 is processed 〉

The bit string of picture P13 is input to bit string analytic unit 201.Bit string analytic unit 201 carries out extracting from the input bit string processing of various data.Various data are: the expression mode of carrying out is selected the information of usefulness, the i.e. information of presentation code mode Ms (following also referred to as coded system); The information (following also referred to as motion vector) of expression motion vector MV, heading message, coded data (view data) etc.The coded system Ms that extracts outputs to mode decoding unit 203.In addition, the motion vector MV that extracts outputs to motion compensation decoding unit 205.And then the coded prediction error data Ed that is extracted by bit string analytic unit 201 outputs to predicated error decoding unit 202.

Mode decoding unit 203 based on the coded system Ms that extracts from bit string, carries out the switching controls of switch 209 and 210.In the situation of within coded system represents picture, encoding, switch 209 is controlled so that input terminal Te is connected to lead-out terminal Tf1, and switch 210 is controlled so that lead-out terminal Th is connected to input terminal Tg1.In addition, represent in coded system in the situation of inter-picture prediction coding switch 209 to be controlled so that input terminal Te is connected to lead-out terminal Tf2, switch 210 is controlled so that lead-out terminal Th is connected to input terminal Tg2.In addition, mode decoding unit 203 outputs to motion compensation decoding unit 205 with coded system Ms.

Below, coded system is represented the situation of inter-picture prediction coding describes.

Predicated error decoding unit 202 is decoded to the coded data Ed that inputs, generation forecast error information PDd.The prediction error data PDd that generates outputs to switch 209.In the case, because the input terminal Te of switch 209 is connected to lead-out terminal Tf2, so prediction error data PDd outputs to add operation unit 208.

Motion compensation decoding unit 205 carries out motion compensation based on the motion vector MV that is extracted by analytic unit 201 with reference to picture index Rp, and obtains motion compensated image from reference picture memory 207.This motion compensated image is corresponding to the image in the zone of object piece to be decoded in the object picture.

Picture P13 utilizes picture P4, P7, P10 to implement coding as forward direction with reference to candidate screen.Decoding for picture P13 when processing, these candidate screen are decoded and be stored in reference in the picture memory 207.

Therefore, motion compensation decoding unit 205 determines be which picture among above-mentioned picture P4, P7, the P10 is used as with reference to picture when the object piece decoding of above-mentioned picture P13.Then, motion compensation decoding unit 205, the based on motion vector from reference picture memory 207, is obtained with reference to the image in the zone of the object piece in the picture correspondence as motion compensated image.

The below illustrates the time dependent situation of picture by reference picture memory 207 storage with Fig. 3, and with reference to definite method of picture.

By memory control unit 204 controls, which kind of this control system extracts based on the heading message from bit string, expression carries out with reference to the information (with reference to image information) that obtains P picture and B picture with reference to picture memory 207.

As shown in Figure 3, the storage area (#1)～(#5) that has 5 picture amounts with reference to picture memory 207.When the decoding of picture P13 begins, in reference picture memory 207, store picture B8, P4, P7, P10, B9.Picture P13 is used as picture P4, P7, P10 to decode with reference to the candidate of picture.In the storage area of having stored picture B8, the picture P13 that storage is decoded.It is the reasons are as follows, picture P4, P7, P10 when picture P13 and later picture thereof are decoded, be used as the candidate screen with reference to picture, and picture B8 are not used as with reference to picture to these picture codecs the time.

In Fig. 3, the picture shown in each circular mark is the moment of having finished in the decoding of object picture, is stored at last the picture (object picture) with reference to picture memory 207.

In the case, when the motion vector of the object piece that detects picture P13, be just can be according to being additional to determining with reference to image information of motion vector with reference to which picture.

Specifically be exactly with reference to picture index with reference to image information, and, to giving with reference to picture index with reference to candidate screen of picture P13.What the reference candidate screen was given is nearer in time apart from object picture (picture P13) with reference to candidate screen with reference to picture index, and value is just less.

Particularly, in the situation with reference to picture P10 when picture P13 object piece is encoded, the information (for example with reference to picture index [0]) that the candidate screen in tight the place ahead of indicated object picture (picture P10) is used as with reference to picture is is just recorded and narrated in the bit string of object piece.In addition, in the situation with reference to picture P7 when the object piece is encoded, the information (for example with reference to picture index [1]) that the candidate screen before 2 of the indicated object picture is used as with reference to picture is is just recorded and narrated in the bit string of this object piece.In addition, in the situation with reference to picture P4 when picture P13 object piece is encoded, the candidate screen before 3 of the indicated object picture is used as the information (for example with reference to picture index [2]) with reference to picture, just records and narrates in the bit string of this object piece.

With reference to picture index, just can understand which candidate screen is used as with reference to picture when the object piece is encoded according to this.

Like this, the motion compensated image image of the zone of corresponding objects piece in the picture (namely with reference to) according to motion vector with reference to image information, from reference picture memory 207, is obtained in motion compensation decoding unit 205.

The motion compensated image that generates like this outputs to add operation unit 208.

In addition, motion compensation decoding unit 205 when carrying out the decoding of P picture, outputs to motion vector storage unit 226 with motion vector MV with reference to image information Rp.

Add operation unit 208 is with data (prediction data) the Pd addition of prediction error data PDd and motion compensated image and generating solution code data Ad.The decoded data Ad that generates outputs to reference to picture memory 207 as decode image data DId by switch 210.

Like this, the piece of picture P13 is decoded in order.When whole pieces of picture P13 are all decoded, just carry out the decoding of picture B11.

The decoding of＜picture B11 is processed 〉

Because the action of bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, identical when processing with the decoding of picture P13 is so the description thereof will be omitted.

Motion compensation decoding unit 205 is from the Information generation moving compensating data of motion vector of inputting etc.What output to motion compensation decoding unit 205 from bit string analytic unit 201 is motion vector and with reference to picture index.By with picture P7, B9 and the P10 candidate screen as the forward direction reference, and picture P13 processed as the predictive coding of the candidate screen of backward reference and obtain picture B11.These have been implemented with reference to candidate screen during decoding processes and be stored in reference to picture memory 207 the object picture being decoded constantly.

Below, with Fig. 3 the time dependent situation of picture by reference picture memory 207 storage is described, and with reference to definite method of picture.

With reference to picture memory 207 by memory control unit 204 control, that this control system extracts based on the heading message from bit string, be illustrated in carried out when P picture and B picture carried out coding which kind of with reference to and so on information Ih.

When the decoding of picture B11 begins, in the reference picture memory, store picture P13, P4, P7, P10, B9.For picture B11, with picture P7, B9, the P10 candidate screen as the forward direction reference, and picture P13 implemented decoding with reference to picture and process as backward.In the storage area of stored picture P4, the picture B11 that decodes of storage, this is because picture P4 is not used as the candidate with reference to picture when picture B11 and later picture thereof are decoded.

In the case, when detecting forward motion vector, be with reference to which candidate screen, just can be according to being additional to determining with reference to image information of motion vector.

Particularly, in the situation with reference to picture P10 when the object piece of picture B11 is encoded, the information (for example with reference to picture index [0]) that the candidate screen in tight the place ahead of indicated object picture (picture P10) is used as with reference to picture is is just recorded and narrated in the bit string of this object piece.In addition, when when the object piece is encoded with reference in the situation of picture B9, the information (for example with reference to picture index [1]) that the candidate screen before 2 of the indicated object picture is used as with reference to picture is is just recorded and narrated in the bit string of this object piece.In addition, when when the object piece is encoded with reference in the situation of picture P7, the candidate screen before 3 of the indicated object picture is used as the information (for example with reference to picture index [2]) with reference to picture, just records and narrates in the bit string of this object piece.

Thereby, with reference to picture index, just can understand which candidate screen is used as with reference to picture when the object piece is encoded according to this.

Motion compensation decoding unit 205 is in the situation of bi-directional predictive coding in the mode of selecting, and according to the reference picture index, determines to use which picture of picture P7, B9 and P10 to carry out the forward direction reference.Then, motion compensation decoding unit 205 is obtained the forward motion compensation image based on forward motion vector from reference picture memory 207, in addition, obtains the reverse compensating images based on backward motion vector from reference picture memory 207.

Then, motion compensation decoding unit 205 by the summation averaging of forward motion compensation image and reverse compensating images, generates motion compensated image.

Then, the processing of using the forward and backward motion vector to generate motion compensated image is described.

(bi-directional predicted mode)

Figure 17 represents that the decoder object picture is picture B11, and piece (object piece) BLa01 of the decoder object that becomes picture B11 is carried out the situation of bi-directional predicted decoding.

At first, illustrate that forward direction is picture P10 with reference to picture, and backward be the situation of picture P13 with reference to picture.

In the case, forward motion vector is the motion vector MVe01 of expression picture regional CRe01 P10, corresponding blocks BLa01.Backward motion vector is expression, the motion vector MVg01 of regional CRg01 picture P13, corresponding blocks BLa01.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain the image of the regional CRe01 among the picture P10 with reference to image as forward direction, as the backward image of obtaining the regional CRg01 among the picture P13 with reference to image, the image of these two regional CRe01 and CRg01 is implemented the summation averaging of view data, obtain the motion compensated image corresponding to above-mentioned object piece BLa01.

Then, illustrate that forward direction is picture B9 with reference to picture, backward is the situation of picture P13 with reference to picture.

In the case, forward motion vector is the motion vector MVf01 of the regional CRf01 of the corresponding blocks BLa01 of expression picture B9.Backward motion vector is the motion vector MVg01 of the regional CRg01 of the corresponding blocks BLa01 of expression picture P13.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain the image of the regional CRf01 among the picture B9 with reference to image as forward direction, as the backward image of obtaining the regional CRg01 among the picture P13 with reference to image, the image of these two regional CRf01 and CRg01 is implemented the summation averaging of view data, obtain the motion compensated image corresponding to above-mentioned object piece BLa01.

(direct mode)

In addition, be in the situation of direct mode in coded system, rear motion vector (benchmark motion vector) to reference picture relative position P13, picture the piece identical with the object piece that store, object picture B11 is obtained in motion compensation decoding unit 205 in motion vector storage unit 226.Motion compensation decoding unit 205 uses this benchmark motion vector, obtains forward direction with reference to image and backward with reference to image from reference picture memory 207.Then, the 205 pairs of forward directions in motion compensation decoding unit are with reference to image and backward summation averaging with reference to image enforcement view data, thereby generation is corresponding to the motion compensated image of above-mentioned object piece.In the following description, for the piece in the relative position of picture and other pictures that specific piece in the picture equates, also be called simply the piece that is in specific same position of a picture.

Figure 18 (a) expression is with reference near the picture P10 before the picture B11, the situation (the 1st example of direct mode decoding) of in direct mode the piece BLa10 of picture B11 being decoded.

The benchmark motion vector that uses in the decoded process under direct mode at piece BLa10 is, when piece BLa10 is decoded after to picture (reference pictures) P13 of reference, be in forward motion vector (benchmark motion vector) MVh10 with piece (reference block) BLg10 of piece BLa10 same position.This forward direction motion vector MVh10, expression near the picture P10 before the picture B11, corresponding to the motion vector of the regional CRh10 of reference block BLg10.

In the case, as the forward motion vector MVk10 of piece BLa10 to be decoded, use motion vector regional CRk10, parallel with said reference motion vector MVh10 among the above-mentioned picture P10 of expression, corresponding objects piece BLa10.In addition, among the backward motion vector MVi10 as the piece BLa10 of object to be decoded, use motion vector regional CRi10, parallel with said reference motion vector MVh10 among the above-mentioned picture P13 of expression, corresponding objects piece BLa10.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image among the regional CRk10 of picture P10 as forward direction with reference to image, obtain backward with reference to the image among the regional CRi10 of picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) corresponding to object piece BLa10.

In the case, the size (MVB) of the size of forward motion vector MVk10 (MVF) and backward motion vector MVi10 can be used the size (MVR) of said reference motion vector MVh10, is obtained by above-mentioned formula (1), (2).

If big or small MVF, the MVB of each motion vector represent respectively horizontal component and the vertical component of motion vector.

In addition, TRD is, the object piece BLa10's of picture B11 is backward with reference to picture P13 with to carry out the time gap of the picture P10 of forward direction reference when decoded after this to piece (reference block) BLg10 of reference picture (reference pictures) P13.In addition, TRF is, picture B11 and near the time gap with reference to picture P10 before it, TRB be, picture B11 and the time gap of the picture P10 of reference when backward piece BLg10 with reference to picture P13 decoded.

Figure 18 (b) expression is with reference near the picture P10 before the picture B11, the situation (the 2nd example of direct mode decoding) of in direct mode the piece BLa10 of picture B11 being decoded.

In the 2nd example of this direct mode decoding, different from the 1st example of the direct mode decoding shown in Figure 18 (a), picture P7 at the picture that reference block (namely for object piece backward with reference to picture, with the piece of object piece same position) is carried out forward direction reference when decoded.

Namely, the benchmark motion vector that uses in the decoded process under direct mode at piece BLa20 is, when piece BLa20 is decoded by rear in the picture P13 of reference, be in the forward motion vector MVh20 with the piece BLg20 of piece BLa20 same position.This forward direction motion vector MVh20, be expression the place ahead of being positioned at object picture B11 picture P7, corresponding to the motion vector of the regional CRh20 of reference block BLg20.

In the case, as the forward motion vector MVk20 of piece BLa20 to be decoded, use motion vector regional CRk20, parallel with said reference motion vector MVh20 among the above-mentioned picture P10 of expression, corresponding objects piece BLa20.In addition, as the backward motion vector MVi20 of piece BLa20 to be decoded, use motion vector regional CRi20, parallel with said reference motion vector MVh20 among the above-mentioned picture P13 of expression, corresponding objects piece BLa20.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image among the regional CRk20 of picture P10 as forward direction with reference to image, obtain backward with reference to the image among the regional CRi20 of picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) corresponding to object piece BLa20.

In the case, the size of forward motion vector MVk20 (MVF), and the size (MVB) of backward motion vector MVi20, can use the size (MVR) of said reference motion vector MVh20, same with the 1st example of above-mentioned direct mode decoding, obtained by above-mentioned formula (1), (2).

Figure 19 (a) expression is with reference to the picture P7 that is positioned near the place ahead of the picture P10 before the picture B11, the situation (the 3rd example of direct mode decoding) of in direct mode the piece BLa30 of picture B11 being decoded.

In the 3rd example of this direct mode decoding, different with the 2nd example from the 1st example of the direct mode decoding shown in Figure 18 (a) and Figure 18 (b), at the picture that the object piece is carried out forward direction reference when decoded, be not near the picture before the object picture, but when the reference block (with the piece of object piece same position) of reference pictures is decoded the picture of forward direction reference.Reference pictures be when the object piece is decoded after to the picture of reference.

Namely, the benchmark motion vector that uses in the decoded process under direct mode at piece BLa30 is, when piece BLa30 is decoded after in the picture P13 of reference, be in the forward motion vector MVh30 with the piece BLg30 of piece BLa30 same position.This forward direction motion vector MVh30 is the motion vector of regional CRh30 picture P7, corresponding reference block BLg30 that expression is positioned at the place ahead of object picture B11.

In the case, as the forward motion vector MVk30 of piece BLa30 to be decoded, use motion vector regional CRk30, parallel with said reference motion vector MVh30 among the above-mentioned picture P7 of expression, corresponding objects piece BLa30.In addition, as the backward motion vector MVi30 of piece BLa30 to be decoded, use expression motion vector regional CRi30 above-mentioned picture P13, corresponding objects piece BLa30, parallel with said reference motion vector MVh30.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of the regional CRk30 of picture P7 as forward direction with reference to image, the image of obtaining backward regional CRi30 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) corresponding to object piece BLa30.

In the case, the size of forward motion vector MVk30 (MVF), and the size (MVB) of backward motion vector MVi30 can be used the size (MVR) of said reference motion vector MVh30, are obtained by above-mentioned formula (2), (3).

When at the picture of piece BLg30 institute's reference of decoded time in the situation of reference picture memory 207 deletions, with the immediate forward direction picture in position P10 in time as the forward direction in the 3rd example of direct mode decoding with reference to picture.Processing in the 3rd example of direct mode decoding in the case, identical with the 1st example of direct mode decoding.

It is 0 motion vector that size is used in Figure 19 (b) expression, the situation (the 4th example that direct mode is decoded) of in direct mode the piece BLa40 of picture B11 being decoded.

In the 4th example of this direct mode decoding, the size of the benchmark motion vector that uses in the 1st example of the direct mode decoding shown in Figure 18 (a) and Figure 18 (b) and the 2nd example is 0.

In the case, in the forward motion vector MVk40 and backward motion vector MVi40 of piece BLa40 to be decoded, the use size is 0 motion vector.

Namely, above-mentioned forward motion vector MVk40, among the expression picture P10, be positioned at object piece BLa40 same position, with zone (piece) CRk40 of the same size of object piece.In addition, above-mentioned backward motion vector MVi40, among the expression picture P13, be positioned at object piece BLa40 same position, with zone (piece) CRi40 of the same size of object piece.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of zone (piece) CRk40 of picture P10 as forward direction with reference to image, the image of obtaining backward regional CRi40 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) for object piece BLa40.The method can be applicable to, and for example, is not have the such situation of piece of motion vector as by the piece of intraframe coding as the backward of picture B11 with reference to the piece that is in piece BLa40 same position among the picture P13 of picture.

The data of the motion compensated image that generates like this are to 208 outputs of add operation unit.In add operation unit 208, the prediction error data inputted and the data of motion compensated image are added up mutually, generate the view data of decoding.The decode image data that generates is output to reference to picture memory 207 by switch 210, and decoded picture is stored in this with reference to picture memory 207.

Memory control unit 204, extract based on the heading message from bit string, P picture and B picture be which kind of adopts with reference to the information Ih that encodes and so on, reference picture memory 207 is controlled.

As above above-mentioned, the piece of picture B11 is sequentially decoded.When whole pieces of picture B11 are decoded, carry out the decoding of picture B12.

In the decoded process of above-mentioned B picture, sometimes specific piece is processed as jumping over piece., below, the decoding of jumping over piece is described simply.

In the decoding of the bit string of inputting is processed, by index information that jumps over identifier and piece of in this bit string, recording and narrating etc., distinguish that specific is used as and jumps in the situation that piece processes, carry out in direct mode motion compensation, namely obtaining corresponding to the predicted picture of object piece.

For example, shown in Fig. 6 (b), with the piece MB (r) of picture B11 and the piece MB (r+1) between the piece MB (r+1) and piece MB (r+2) as jumping in the situation that piece processes, detected by bit string analytic unit 201 and to jump over identifier Sf among the bit string Bs.When this jumped over identifier Sf and is imported into mode decoding unit 223,223 pairs of motion compensation decoding unit of mode decoding unit send instruction made it carry out in direct mode motion compensation.

So, motion compensation decoding unit 205, based on be included in forward direction with reference among the picture P10 and with as the image (forward direction is with reference to image) that jumps over piece that piece processes and be in the piece of same position, with with as the image (backward with reference to image) that jumps over piece that piece processes and be in the piece of same position, obtain the predicted picture of piece MB (r+1) and MB (r+2), and the data of predicted picture are outputed to add operation unit 208.Predicated error decoding unit 202, as the differential data that is treated to the piece that jumps over piece, output valve is 0 data.In add operation unit 208, owing to be 0 as the differential data that jumps over the piece that piece processes, so the data of the predicted picture of piece MB (r+1) and MB (r+2), the data that are used as the decoded picture of piece MB (r+1) and MB (r+2) output to reference to picture memory 207.

In addition, in processing example (the 2nd example) under the direct mode shown in processing under the direct mode shown in Figure 18 (a) example (the 1st example), Figure 18 (b), the processing example (the 3rd example) under the direct mode shown in 19 (a), not to be that 0 piece is all processed as jumping over piece with differential data.In other words, with relative object picture be positioned in time near front picture as forward direction with reference to picture, and be that 0 motion vector carries out bi-directional predicted for the object piece with size, and only under the differential data of object piece is 0 situation, this object piece is processed as jumping over piece.

Under these circumstances, when the identifier etc. of jumping over from bit string Bs is learnt specific piece when jumping over piece, motion compensation process, should by will be in time near front with reference to picture as forward direction with reference to picture, motion is 0 bi-directional predicted carrying out.

The decoding of＜picture B12 is processed 〉

Because the action of bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, identical when processing with the decoding of picture P10 is so the description thereof will be omitted.

Motion compensation decoding unit 205 is from the Information generation motion-compensated image data of motion vector of inputting etc.That be input to motion compensation decoding unit 205 is motion vector MV and with reference to picture index Rp.Picture B12 is with picture P7, P10 and the B11 candidate screen as the forward direction reference, picture P13 implemented to encode as the candidate screen of backward reference process.These candidate screen the object picture is decoded constantly decoded and be stored in reference to picture memory 207.

By the time dependent situation of picture of reference picture memory 207 storage, and with reference to definite method of picture, the method when illustrated picture B11 is decoded with using Fig. 3 is identical.

Be in the situation of bi-directional predictive coding mode in the presentation code mode, motion compensation decoding unit 205 according to the reference picture index, determines which picture of picture P7, P10 and B11 has been carried out the forward direction reference.Then, forward direction is obtained with reference to image based on forward motion vector from reference picture memory 207 in motion compensation decoding unit 205, and, in addition, obtain backward with reference to image from reference picture memory 207 based on backward motion vector.Then, motion compensation decoding unit 205, by to forward direction with reference to image with backwardly implement the summation averaging of view data, the motion compensated image of formation object piece with reference to image.

(bi-directional predicted mode)

Figure 20 represents that the decoder object picture is picture B12, and piece (object piece) BLa02 of the decoder object that becomes picture B12 is carried out the situation of bi-directional predicted decoding.

At first, illustrate that forward direction is picture B11 with reference to picture, backward is the situation of picture P13 with reference to picture.

In the case, forward motion vector is the motion vector MVe02 corresponding to the regional CRe02 of piece BLa02 of expression picture B11.Backward motion vector is the motion vector MVg02 corresponding to the regional CRg02 of piece BLa02 of expression picture P13.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, the image of obtaining the regional CRe02 among the picture B11 as forward direction with reference to image, the image of obtaining the regional CRg02 among the picture P13 as backward with reference to image, the image of these two regional CRe02 and CRg02 is implemented the summation averaging of view data, obtain the motion compensated image for above-mentioned object piece BLa02.

Then, be picture P10 to forward direction with reference to picture, backward is that the situation of picture P13 is explained with reference to picture.

In the case, forward motion vector is the motion vector MVf02 corresponding to the regional CRf02 of piece BLa02 of expression picture P10.Backward motion vector is the motion vector MVg02 corresponding to the regional CRg02 of piece BLa02 of expression picture P13.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, the image of obtaining the regional CRf02 among the picture P10 as forward direction with reference to image, the image of obtaining the regional CRg02 among the picture P13 as backward with reference to image, the image of these two regional CRf02 and CRg02 is implemented the summation averaging of view data, obtain the motion compensated image of above-mentioned object piece BLa02.

(direct mode)

In addition, be in the situation of direct mode in coded system, the motion vector (benchmark motion vector) of rear reference block to reference picture P13 that store, object picture B12 (piece that relative position is identical with the object piece) is obtained in motion compensation decoding unit 205 in motion vector storage unit 226.Forward direction by using this benchmark motion vector, is obtained with reference to image and backward with reference to image from reference picture memory 207 in motion compensation decoding unit 205.Then, the 205 pairs of forward directions in motion compensation decoding unit generate the motion compensated image for above-mentioned object piece with reference to image and backward with reference to image enforcement view data summation averaging.

Figure 21 (a) expression is with reference near the picture B11 before the picture B12, the situation (the 1st example of direct mode decoding) of in direct mode the piece BLa50 of picture B12 being decoded.

The benchmark motion vector that uses in the decoded process under direct mode at piece BLa50 is, when piece BLa50 is decoded after to the forward motion vector MVj50 of the reference block (being in the piece BLg50 with piece BLa50 same position) of the picture P13 of reference.This forward direction motion vector MVj50, be expression the place ahead picture P10 nearby of being positioned at picture B11, corresponding to the motion vector of the regional CRj50 of reference block BLg50.

In the case, as the forward motion vector MVk50 of the piece BLa50 of decoder object, use expression be included in above-mentioned picture B11, corresponding to the motion vector regional CRk50, parallel with said reference motion vector MVj50 of object piece BLa50.In addition, as the backward motion vector MVi50 of the piece BLa50 of decoder object, use expression be included in above-mentioned picture P13, corresponding to the motion vector regional CRi50, parallel with said reference motion vector MVj50 of object piece BLa50.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of the regional CRk50 of picture B11 as forward direction with reference to image, the image of obtaining backward regional CRi50 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) for object piece BLa50.

In the case, the size of forward motion vector MVk50 (MVF), and the size (MVB) of backward motion vector MVi50 can be used the size (MVR) of said reference motion vector MVj50, are obtained by above-mentioned formula (1), (2).

Big or small MVF, the MVB of each motion vector be horizontal component and the vertical component of expressive movement vector respectively.

Figure 21 (b) expression is with reference to the picture B11 in the place ahead of picture B12, the situation (the 2nd example of direct mode decoding) of in direct mode the piece BLa60 of picture B12 being decoded.

In the 2nd example of this direct mode decoding, different from the 1st example of direct mode shown in Figure 21 (a) decoding, reference block (namely the object piece backward with reference in the picture, and the piece of object piece same position) picture of forward direction reference when decoded is picture P7.

Namely, the benchmark motion vector that uses in the decoded process under direct mode at piece BLa60 is, when piece BLa60 is decoded after to the forward motion vector MVj60 of the reference block (being in the piece BLg60 with piece BLa60 same position) of the picture P13 of reference.This forward direction motion vector MVj60, be expression the place ahead of being positioned at object picture B12 picture P7, corresponding to the motion vector of the regional CRj60 of reference block BLg60.

In the case, as the forward motion vector MVk60 of the piece BLa60 of decoder object, use the above-mentioned picture B11 of expression, corresponding to the motion vector regional CRk60, parallel with said reference motion vector MVj60 of object piece BLa60.In addition, as the backward motion vector MVi60 of the piece BLa60 of decoder object, use the above-mentioned picture P13 of expression, corresponding to the motion vector regional CRi60, parallel with said reference motion vector MVj60 of object piece BLa60.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of the regional CRk60 of picture B11 as forward direction with reference to image, the image of obtaining backward regional CRi60 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) of object piece BLa60.

In the case, the size of forward motion vector MVk60 (MVF), and the size (MVB) of backward motion vector MVi60, can use the size (MVR) of said reference motion vector MVj60, the same with the 1st example of above-mentioned direct mode decoding, obtained by above-mentioned formula (1) and (2).

Figure 22 (a) expression is with reference to being positioned at than at the immediate forward direction picture P10 of the picture B12 picture P7 in the place ahead more, the situation (the 3rd example of direct mode decoding) of in direct mode the piece BLa70 of picture B12 being decoded.

In the 3rd example of this direct mode decoding, different with the 2nd example from the 1st example of the direct mode decoding shown in Figure 21 (a) and Figure 21 (b), the picture of forward direction reference when the object piece is decoded, be not near the picture before the object picture, but when the reference block of reference pictures is decoded the picture of forward direction reference.Reference pictures be when the object piece is decoded after to the picture of reference.

Namely, the benchmark motion vector that uses in the decoded process under direct mode at piece BLa70 is, when piece BLa70 is decoded after to the forward motion vector MVj70 of reference block (being in the piece with the piece BLa70 same position) BLg70 of the picture P13 of reference.This forward direction motion vector MVj70, be expression the place ahead of being positioned at object picture B12 picture P7, corresponding to the motion vector of the regional CRj70 of reference block BLg70.

In the case, as the forward motion vector MVk70 of the piece BLa70 of decoder object, use the above-mentioned picture P7 of expression, corresponding to the motion vector regional CRk70, parallel with said reference motion vector MVj70 of object piece BLa70.In addition, as the backward motion vector MVi70 of the piece BLa70 of decoder object, use the above-mentioned picture P13 of expression, corresponding to the motion vector regional CRi70, parallel with said reference motion vector MVj70 of object piece BLa70.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of the regional CRk70 of picture P7 as forward direction with reference to image, the image of obtaining backward regional CRi70 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) of object piece BLa70.

In the case, the size of forward motion vector MVk70 (MVF), and the size (MVB) of backward motion vector MVi70 can be used the size (MVR) of said reference motion vector MVj70, are obtained by above-mentioned formula (2), (3).

When at the picture of piece BLg70 institute's reference of decoded time in the situation of reference picture memory 207 deletions, with the immediate forward direction picture in position P10 in time as the forward direction in the 3rd example of direct mode decoding with reference to picture.Processing in the 3rd example of direct mode decoding in the case, identical with the 1st example of direct mode decoding.

It is 0 motion vector that size is used in Figure 22 (b) expression, the situation (the 4th example that direct mode is decoded) of in direct mode the piece BLa80 of picture B12 being decoded.

In the 4th example of this direct mode decoding, be 0 with the 1st example of the decoding of the direct mode shown in Figure 21 (a) and Figure 21 (b) and the size of the benchmark motion vector in the 2nd example.

In the case, as forward motion vector MVk80 and the backward motion vector MVi80 of the piece BLa80 of decoder object, using size is 0 motion vector.

Namely, above-mentioned forward motion vector MVk80, expression picture B11, be positioned at object piece BLa80 same position, with zone (piece) CRk80 of the same size of object piece.In addition, above-mentioned backward motion vector MVi80, expression picture P13, be positioned at object piece BLa80 same position, with zone (piece) CRi80 of the same size of object piece.

Thereby, motion compensation decoding unit 205, from reference picture memory 207, obtain forward direction with reference to the image of zone (piece) CRk80 of picture B11 as forward direction with reference to image, the image of obtaining backward regional CRi80 with reference to picture P13 as backward with reference to image, carry out summation averaging by the view data for these two images, obtain the motion compensated image (predicted picture) for object piece BLa80.The method also can be applied to, and for example, is not have the such situation of piece of motion vector as by the piece of intraframe coding as the backward of picture B12 with reference to the piece that is in piece BLa80 same position among the picture P13 of picture.

The data of the motion compensated image that generates like this output to add operation unit 208.Add operation unit 208 is with the data addition of the prediction error data inputted and motion compensated image and generate decode image data.The decode image data that generates outputs to reference to picture memory 207 by switch 210.

As mentioned above, the piece of picture B12 being implemented decoding in order processes.The view data of each picture of storing in reference picture memory 207 is rearranged in chronological order and exports as output image data Od afterwards.

Then, shown in Figure 16 (a), for the later picture of picture B12 of arranging by the decode time order, according to picture type, by decoding with the same processing of above-mentioned picture P13, B11, B12.Figure 16 (b) expression is by the arrangement of displaying time row's that order changes picture.

In the decoding of the bit string of inputting is processed, when by index information that jumps over identifier and piece of in this bit string, recording and narrating etc., distinguish that specific is used as and jumps in the situation that piece processes, similarly carry out in direct mode motion compensation with the situation of the decoding of picture B11, namely obtaining corresponding to the predicted picture of object piece.

As mentioned above, because in the moving image decoding apparatus 20 of present embodiment 2, when the piece of B picture is decoded, candidate screen as the forward direction reference is used the B picture of decoding with the P picture of decoding, based in the bit string corresponding to object piece to be decoded, comprise, when being illustrated in this object piece and being encoded by the information of the candidate screen of forward direction reference (with reference to picture index), generate the predicted picture of this object piece, so just can be to the B picture is correctly decoded as the piece of the coded B picture of the candidate screen of forward direction reference.

In addition, in moving image decoding apparatus 20, in the situation that the object piece to be decoded of B picture is encoded in direct mode, based on being in the motion vector of the piece of same position with this object piece, come the motion vector of calculating object piece.So namely necessity gets the information of the motion vector of coded in direct mode piece from the coding side-draw in the decoding side.

In addition, in moving image decoding apparatus 20, based on comprise in bit string as heading message, the information of used candidate screen when being illustrated in P picture and B picture and being encoded, manage the data of the decoded picture of in the reference picture memory, storing.For example, in the moment that the decoding of a picture has been finished dealing with, deleting successively in the later decoding of this picture is processed can be as the data with reference to the picture of picture, so just can efficient utilize well picture memory.

In addition, when the object piece of P picture is decoded, which picture in the candidate screen is used as with reference to picture, (namely, the motion vector of object piece to be decoded with reference to which candidate screen obtains) can determining with reference to image information from be attached to motion vector information.

Equally, when the object piece of B picture is decoded, which picture in the candidate screen of forward direction reference is used as with reference to picture, (namely, the forward motion vector of object piece to be decoded with reference to which candidate screen obtains) can determining with reference to image information from be attached to motion vector information.

Although in present embodiment 2, what adopt for one of a plurality of coded systems of B picture is direct mode, also can not use direct mode as the coded system of B picture.In the case, the motion vector storage unit 226 that does not just need moving image decoding apparatus 20.

In addition, in present embodiment 2, as direct mode, 4 methods (namely the 4th example shown in the 3rd example shown in the 2nd example shown in the 1st example shown in Figure 18 (a) or Figure 21 (a), Figure 18 (b) or Figure 21 (b), Figure 19 (a) or Figure 22 (a), Figure 19 (b) or Figure 22 (b)) been have specifically have been recorded and narrated, in decoding device, use and in code device, decode as method corresponding to the used method of direct mode.Specifically, using as direct mode in the situation of a plurality of methods, use is recorded and narrated in bit string, and expression has used the information of which kind of method to decode as concrete direct mode.

At this moment, the action of motion compensation decoding unit 205 just changes according to this information.For example, add in the situation of this information as unit at the piece take motion compensation, determine the codings that used which method in above-mentioned 4 concrete grammars to carry out as direct mode by mode decoding unit 223, and the concrete grammar of determined direct mode is sent to motion compensation decoding unit 205.Then, motion compensation decoding unit 205, which is used in above-mentioned 4 methods according to the concrete grammar as direct mode, carries out suitable motion compensated prediction decoding and processes.

In addition, the information (DM mode information) that is used and so in which method of concrete grammar as direct mode, record and narrate in title, GOP title, picture title or the situation as in the bar title in sequence integral body, to each sequence integral body, GOP, picture or picture bar, DM mode information is sent to motion compensation decoding unit 205 from bit string analytic unit 201, the action of motion compensation decoding unit 205 is changed.

Although in present embodiment 2,2 B pictures are between I picture and the P picture or between the P picture of adjacency, and the number of continuous B picture also can be 3 or 4.

In addition, although in present embodiment 2, the quantity of the candidate screen of the forward direction reference of adopting for the P picture is that the quantity with reference to candidate screen of 3, P picture also can be the value beyond 3.

In addition, although in present embodiment 2, the candidate screen of the forward direction reference of using when the B picture is decoded is 2 I or P picture and 1 B picture, and the candidate screen of forward direction reference is not limited to this when the B picture is decoded.

In addition, in present embodiment 2, as the method that when picture P13, picture B11 and picture B12 are decoded, the reference picture memory is managed, as shown in Figure 3, to as being illustrated with reference to the P picture of the candidate of picture and the method that the B picture manages in the lump.But with reference to the management method of picture memory, also can be 4 the illustrated methods of use Figure 11～Figure 14 in execution mode 1, namely, will be used as whole pictures with reference to the candidate of picture and be divided into the method that P picture and B picture manage.

In the case, with reference to picture memory 207, the zone that has 6 picture amounts as storage area, namely, P picture memory (#1)～(#4), B picture memory (#1) and (#2).In addition, these 6 storage areas are not limited to be formed on 1 with reference to the situation in the picture memory, also can be that 6 storage areas independently consist of with reference to picture memory by 1 respectively.

In addition, as shown in figure 14, when the coding side, to each picture to be encoded, which is preferentially given in the situation with reference to the method for picture index with determining to make P picture storage area and B picture storage area, moving image decoding apparatus can be illustrated in information in the bit string that is in preferential storage area in above-mentioned two storage areas by use, is identified in simply the picture that is used as in the candidate screen with reference to picture based on the reference picture index.

For example, be in the situation of picture B11 at the decoder object picture since apart from this object picture in time hithermost forward direction be picture P10 with reference to picture, so that be stored in picture in the P picture memory and preferentially carry out index and give.Thereby, when the object piece of picture B11 is encoded, as heading message, additional with reference to picture index [0] under picture P10 is used as with reference to the situation of picture in the bit string of object piece.Similarly, additional with reference to picture index [1] under picture P7 is used as with reference to the situation of picture, and additional with reference to picture index [2] under picture B9 is used as with reference to the situation of picture.Thereby moving image decoding apparatus just can be according to the reference picture index, and understanding which candidate screen when the object piece is encoded is used as with reference to picture.

In the case, because expression makes the candidate screen of P picture memory preferentially carry out the information that index is given, be used as heading message and be included in the bit string, so just can by utilizing this information, more easily identify with reference to picture.

In addition, be in the situation of picture B12 at the decoder object picture since apart from this object picture in time hithermost forward direction be picture B11 with reference to picture, so that be stored in picture in the B picture memory and preferentially carry out index and give.Thereby, when the object piece of picture B12 is encoded, as heading message, additional with reference to picture index [0] under picture B11 is used as with reference to the situation of picture in the bit string of object piece.Similarly, additional with reference to picture index [1] under picture P10 is used as with reference to the situation of picture, and additional with reference to picture index [2] under picture P7 is used as with reference to the situation of picture.Thereby moving image decoding apparatus just can be according to the reference picture index, and understanding which candidate screen when the object piece is encoded is used as with reference to picture.

In the case, because expression makes the candidate screen of B picture memory preferentially carry out the information that index is given, be used as heading message and be included in the bit string, so just can by utilizing this information, more easily identify with reference to picture.

In addition, in the coding side, as the management method of reference picture memory, have to use in above-mentioned 5 methods and (with reference to the situation of a method of selecting in advance among Fig. 3, Figure 11～Figure 14), perhaps switch the situation of using a plurality of methods in 5 methods.For example, in the situation that the above-mentioned a plurality of methods of switching manage the reference picture memory, moving image decoding apparatus just can based on the information of recording and narrating in bit string, represent each picture has been used which management method, be determined with reference to picture index.

In addition, in above-mentioned execution mode 2, the quantity with reference to candidate screen of having recorded and narrated for the P picture is 3, is 5 examples with reference to the management method of picture memory (Fig. 3, Figure 11～Figure 14) of 2 P pictures and 1 B picture for the candidate screen of the forward direction reference of B picture.But, also can be applied to respectively with reference to the different situation of the quantity of candidate screen and execution mode 2 with reference to 5 examples of the management method of picture memory.With reference to the quantity of candidate screen in the different situation, also different from execution mode 2 with reference to the capacity of picture memory like this.

In addition, in above-mentioned execution mode 2, be illustrated in being distinguished into reference to candidate screen in the method (4 examples shown in Figure 11～Figure 14) that P picture and B picture manage the reference picture memory, store the P picture into P picture storage area, the B picture is stored into the situation of B picture storage area.But, in the memory cell of stored picture, also can utilize defined short period picture memory in H.263++, and the long period picture memory.For example, have short period picture storage is utilized as P picture storage area, the method that the long period picture memory is utilized as B picture storage area.

[execution mode 3]

Figure 23 is the block diagram that explanation utilizes the dynamic image encoding device of embodiments of the present invention 3.

The dynamic image encoding device 30 of this execution mode 3, can be according to the control signal from the outside, to give method with reference to picture index to candidate screen, switch to according to the method for giving (default adding method) of being undertaken by the rule of initial setting with reference to picture index, with utilize default adding method candidate screen given with reference to picture index, and then, according to any one to the adaptation adding method that changes with reference to picture index given of coding situation with adapting to.

Particularly, the action of an action form of the dynamic image encoding device 30 of this execution mode 3 dynamic image encoding device 10 that is execution mode 1.In other words, the dynamic image encoding device 30 of execution mode 3, select default adding method as its situation with reference to the adding method of picture index under, then carry out the processing same with the dynamic image encoding device 10 of execution mode 1.

Below, dynamic image encoding device 30 is described in detail.

This dynamic image encoding device 30 is with the coding-control unit 110 in the dynamic image encoding device 10 of coding-control unit 130 replacement execution modes 1.Coding-control unit 130, according to the control signal Cont from the outside, to give method with reference to picture index to candidate screen, switch to according to being given the method with reference to picture index (default adding method) by the rule of initial setting to candidate screen, utilize default adding method to give the 1st step with reference to picture index to candidate screen with comprising, with any one of the method (adaptation adding method) of the 2nd step that changes with reference to picture index of giving candidate screen to utilizing default adding method with adapting to.

In addition, coding-control unit 130 has, by each coded object picture to a plurality of with reference to candidate screen, detect respectively as the detecting unit (not shown) with reference to the code efficiency in the picture situation.This coding-control unit 130, according to by the detected code efficiency of this detecting unit, change for each candidate screen of utilizing default adding method to set with reference to picture index.

More specifically, 130 pairs of coding-control unit utilize the value with reference to picture index of each candidate screen that default adding method sets, so that for object picture a plurality of with reference to candidate screen in, as higher with reference to the code efficiency of the object picture in the picture situation, less with reference to the value of picture index.

Then, mode selected cell 139, in direct mode, as for the forward direction of object piece with reference to picture, the picture and selecting that will be endowed with reference to picture index [0] is that forward direction is with reference to picture.Mode selected cell 139, the predictive coding mode beyond direct mode for example in the bi-directional predictive coding mode, carries out from the processing of a plurality of candidate screen selections with reference to picture according to code efficiency.

Other assemblies of the dynamic image encoding device 30 of this execution mode 3, identical with in the dynamic image encoding device 10 of execution mode 1.

Then, the action of dynamic image encoding device 30 described.

In this dynamic image encoding device 30, at the control signal Cont of basis from the outside, as the method with reference to picture index of giving for candidate screen, in the selecteed situation of default adding method, the action of dynamic image encoding device 30, identical with the action of the dynamic image encoding device 10 of execution mode 1.

On the other hand, at the control signal Cont of basis from the outside, as the method with reference to picture index of giving for candidate screen, adapt in the selecteed situation of adding method, in this dynamic image encoding device 30, the dynamic image encoding device 10 by the 1st step and execution mode 1 carries out giving with reference to picture index in the same way.

Adapting in the selecteed situation of adding method, in this dynamic image encoding device 30, by the 2nd step, to utilizing changing with reference to picture index that default adding method gives with adapting to.

Below, to adapt in the selecteed situation of adding method, describe with reference to the concrete adding method of picture index.But in the following description, the suppose object picture is picture B12.

At first, in the 1st step, as shown in Figure 3, the candidate screen of forward direction reference is given with reference to picture index, so that less with reference to the value of picture index apart from the nearer candidate screen of object picture.Namely, be endowed with reference to picture index [1], be endowed with reference to picture index [0], be endowed with reference to picture index [2] with reference to picture P7 with reference to picture B11 with reference to picture P10.

Then, in the 2nd step, as shown in figure 24, with reference to picture P10 be altered to [0] with reference to picture index [1], be altered to [1] with reference to picture B11 with reference to picture index [0].

This rewriting with reference to picture index to each object picture is based on code efficiency and carries out.In addition, dynamic image encoding device 30 will represent according to external control signal Cont, be which the information of setting default adding method and adapting to adding method is inserted in the bit string as heading message as adding method.In addition, setting in the situation that adapts to adding method, be that the information of how to carry out also is inserted in the bit string as heading message with reference to giving of picture index.

As mentioned above, in this execution mode 3, just can be according to code efficiency, will under direct mode, should carry out the forward direction reference candidate screen be made as [0] with reference to picture index.

Namely, in execution mode 1 because nearest with reference to candidate screen in time to distance object picture, the value of giving little with reference to picture index, so in direct mode, can only be with reference to distance object picture B12 nearest picture B11 in time.But in present embodiment 3, if code efficiency can improve, then can forward direction with reference to the distance object picture B12 picture beyond the nearest picture B11 in time.

In addition, in the case, be made as P10 rather than B11 owing to this picture B12 can be carried out in direct mode the picture of reference, so just there is no need picture B11 is decoded.Thereby, just can shown in Figure 25 (a), not decode like that and process near the B picture after the P picture, just can in the unwanted situation of picture B11, obtain the high speed that decoding is processed.In addition, even owing to wait the data shortcoming of picture B11 also can decode because transmitting mistake, so can make the Reliability Enhancement of decoding.

As mentioned above, if can at random give with reference to picture index, determine to carry out the picture of reference under direct mode, the picture ground of then can not decode predetermined like that shown in Figure 25 (a) is processed with having a mind.

And then, shown in Figure 25 (b), process even in the situation that 3 B pictures are arranged between the P picture, also can not encode predetermined picture.Therefore, when in advance at the unwanted picture of coding side understanding user, just can omit such picture to obtain the reduction in processing time in the decoding.

In Figure 25 (b), even picture B3 is not decoded, also can carry out the decoding of other pictures.

Namely, in the adding method of execution mode 1 since picture B4 under direct mode with reference to picture B3, so just need to decode to picture B3 for decoded picture B4.But in present embodiment 3, owing to can at random be set in the picture that carries out reference under the direct mode, the decoding of picture B3 just becomes and there is no need thus.

And then, in this execution mode 3, since carry out with reference to picture index give so that apart from the nearer candidate screen of object picture give less value with reference to picture index, and based on the reference picture index determine under the direct mode with reference to picture, so can improve code efficiency by the shortening of motion vector, can accomplish reduction in processing time simultaneously.

In addition, in the decoding side, because when in direct mode the object piece being processed, the forward direction of giving with reference to picture index [0] is used as with reference to picture at once with reference to candidate screen, so just can shorten the time of decoding.

In addition, although in above-mentioned execution mode 3, be based on the candidate screen that code efficiency determines to be rewritten into reference to picture index [0], also can be with the picture of easy reference, for example in time nearest P picture be made as [0] with reference to picture index.

In addition, be that the present invention is not limited to this with reference to the picture of picture index for [0] although in above-mentioned execution mode 3, carry out in direct mode the picture of reference.Such as, also can encode to learning the information of carrying out in direct mode the situation of reference, and based on the information of under this direct mode, carrying out reference, when direct mode, decode.

[execution mode 4]

Figure 26 is the block diagram that explanation utilizes the moving image decoding apparatus 40 of embodiments of the present invention 4.

This moving image decoding apparatus 40, reception is from the bit string of the dynamic image encoding device 30 of execution mode 3, and based in this bit string, comprise, which of above-mentioned default adding method and above-mentioned adaptation adding method expression should adopt with and carry out the information of giving (adding method indication information) with reference to picture index, carry out the decoding of each picture and process.

Namely, the action of an action form of the moving image decoding apparatus 40 of this execution mode 4 moving image decoding apparatus 20 that is execution mode 2.In other words, the moving image decoding apparatus 40 of execution mode 4 is using in the situation of default adding method as the adding method with reference to picture index, and the action of moving image decoding apparatus 40 is identical with the moving image decoding apparatus 20 of execution mode 2.

Below, dynamic image encoding device 40 is described in detail.

This moving image decoding apparatus 40, with the memory control unit 204 in the moving image decoding apparatus 20 of memory control unit 244 replacement execution modes 2, memory control unit 244, based on the adding method indication information that in bit string, comprises as heading message, carry out and the management of above-mentioned default adding method with the some corresponding memories that adapt to adding method.

Other assemblies of the moving image decoding apparatus 40 of this execution mode 4, identical with in the moving image decoding apparatus 20 of execution mode 2.

Then the action of dynamic image encoding device 40 described.

This moving image decoding apparatus 40 carries out moving accordingly with the adding method indication information that comprises as heading message in from the bit string of dynamic image encoding device 30.

Namely, this moving image decoding apparatus 40, in situation about being selected as at coding side default adding method with reference to the adding method of picture index, namely, in bit string, comprise in the situation of expression default adding method selecteed information, similarly move with the moving image decoding apparatus 20 of execution mode 2.

On the other hand, this moving image decoding apparatus 40 is in situation about being selected as at coding lateral adaptation adding method with reference to the adding method of picture index, namely, in bit string, comprise in the situation of the selecteed information of expression adaptation adding method, carry out moving accordingly with the adaptation adding method.In the case, owing to also be included in the bit string as heading message with reference to the information of how to carry out of giving of picture index, so carry out giving of cross index as the basis take this information.

Below, the action that adapts to dynamic image encoding device 40 in the selecteed situation of adding method is described.

In reference picture memory 207, as shown in figure 24, when the object picture is processed, all be stored in the rewriting with reference to candidate screen in each storage area.

Specifically, be in the situation of picture B12 at the decoder object picture, process for the decoding as the picture B12 of this object piece, select to carry out with reference to picture from candidate screen according to the heading message of object piece.

For example, coded system at the object piece is in the situation of bi-directional predictive coding mode, the decoding of object piece is treated to, as forward direction with reference to picture from candidate screen P10, B11, P7, select to be endowed with in the heading message of object piece, comprise with reference to the identical picture with reference to picture index of picture index.In the heading message that is contained in the object piece with reference to picture index in the situation of [1], then as forward direction with reference to picture and selecting candidate screen B11.Then, process for the bi-directional predicted decoding of object piece, as forward direction with reference to picture with reference to candidate screen B11, carry out with reference to picture P13 with reference to picture as backward.

In addition, be in the situation of direct mode in the coded system of object piece, from candidate screen P7, P10, B9, be endowed with reference to the candidate screen (picture P10) of picture index for [0] with reference to picture and selecting as forward direction.Then, process for the decoding of object piece, as forward direction with reference to picture with reference to candidate screen P10, carry out with reference to picture P13 with reference to picture as backward.

As mentioned above, in this execution mode 4, reference picture memory 207 is carried out the management of memory as shown in figure 24, namely, storage management is, uses to have changed the value with reference to picture index of being given by default adding method according to the coding situation and carry out as the value with reference to picture index of each candidate screen.So, just can realize according to code efficiency, rewrite the coding/decoding method corresponding to coding method with reference to picture index of candidate screen.

Namely, in execution mode 2, because nearest with reference to candidate screen in time to distance object picture, the value of giving little with reference to picture index, so in direct mode, can only be with reference to distance object picture B12 nearest picture B11 in time, but in present embodiment 4, can forward direction with reference to the distance object picture B12 picture beyond the nearest picture B11 in time.

In addition, in the case, be P10 rather than B11 owing to carry out the picture of reference when the piece of picture B12 decoded under direct mode, so just there is no need picture B11 is decoded.Thereby, just can shown in Figure 25 (a), not decode like that and process near the B picture after the P picture, just can in the unwanted situation of picture B11, obtain the high speed that decoding is processed.In addition, even owing to wait the data shortcoming of picture B11 also can decode because transmitting mistake, so can make the Reliability Enhancement of decoding.

As mentioned above, if can be chosen as arbitrarily value according to the value with reference to picture index that the coding situation will be given with reference to candidate screen, determine wittingly under direct mode, to carry out the picture of reference, then can shown in Figure 25 (a), not decode predetermined picture like that and process.

And then, shown in Figure 25 (b), process even in the situation that 3 B pictures are arranged between the P picture, also can not encode predetermined picture, when in advance at the unwanted picture of coding side understanding user, just can omit such picture to obtain the reduction in processing time in the decoding.

In Figure 25 (b), even picture B3 is not decoded, also can carry out the decoding of other pictures.

Namely, owing in execution mode 2, picture B4 is decoded with reference to picture B3 under direct mode, so just need to decode to picture B3.But in present embodiment 4, when in the coding side, carry out in direct mode just can picture B3 not being decoded in the situation that the picture of reference can at random set.

In addition, when when the decoding side is processed the object piece in direct mode, be used as with reference to picture at once with reference to candidate screen owing to compose sub forward direction with reference to picture index [0], so just can shorten the time of decoding.

Although in above-mentioned each form of implementation 1～4, be illustrated in the coding of P picture or when decoding not with reference to the situation of B picture, also can be with reference to the B picture when the coding of P picture or decoding.

In addition, although in above-mentioned each form of implementation 1～4, the situation of calculating the time gap between picture based on the displaying time of each picture is shown, also can obtains time gap between picture based on the information beyond the displaying time equal time of each picture.

For example, also can set the Counter Value that just increases progressively whenever picture is processed to each picture, can be based on the time gap between this Counter Value calculating picture.

Particularly, all comprise in the situation of temporal information at video flowing and the audio stream of a content because the unit of temporal information is very little, based on this time information video information and audio-frequency information are managed so that two kinds of Information preservations synchronously and be not easy.But by being managed by the arrangement of Counter Value to above-mentioned each picture, consider that the synchronous management of video information and audio-frequency information just becomes simple.

In addition, in above-mentioned each form of implementation 1～4, with the data unit of processing, for example, make the heading message of GOP and picture etc. and data portion not discretely, be included in the bit string of corresponding each data processing unit and transmit.But they are separated with different stream transmits.

For example, also can be in the situation that data units of transfer such as stream being distinguished into packet transmits, the title section of corresponding above-mentioned picture and data portion are separated transmitted.In the case, also there are title section and data portion not to be included in situation in the same stream.But, in the data of having used packet transmit, even the order that title section and data portion transmit is discontinuous, also just the title section of correspondence transmits with different packets with data portion, corresponding title section and the corresponding relation of data portion are stored by the heading message of packet etc., situation about therefore, being included in the same bit string with title section and data portion in fact is identical.

And then, although in above-mentioned each form of implementation 1～4, above-mentioned with reference to picture index, a plurality of information that are referenced with reference to which picture in the candidate screen are used when being identified in the object piece and being encoded, but this also can be used as with reference to picture index and shows that coding is processed or a plurality of forward directions of the object picture of the object that decoding is processed use with reference to the information of the position of candidate screen for becoming.Particularly, because the adding method with reference to picture index in above-mentioned form of implementation 1 and 2, perhaps in form of implementation 3 and 4 the default adding method, to above-mentioned a plurality of forward directions with reference to candidate screen give apart from the nearlyer value of object picture less with reference to picture index, so just can be according to giving the size with reference to the value of picture index of each forward direction with reference to candidate screen, detect each forward direction with reference to the position of candidate screen (namely, detect each forward direction with reference to candidate screen, whole forward directions with reference to picture in, be which near picture of distance object picture).

In addition, can be in the bit string corresponding to moving image, with the relative position of such each the forward direction candidate screen of expression comprise dividually the position identification information that expression consists of the position of each picture on the displaying time axle of moving image with reference to picture index.This position identification information is different from the temporal information of the displaying time of expression picture, is the information of determining the relative position of each picture.

In addition, in above-mentioned each form of implementation 1～4, be illustrated under the direct mode as reference pictures, use when the piece of the object picture that becomes coding or decoder object is decoded after to the situation of the picture of reference (for the object picture backward with reference to picture).But under direct mode, as reference pictures, can use for the offscreen picture of having processed of the backward reference of object picture, for example, the picture of forward direction reference when the piece of object picture is encoded.

[execution mode 5]

Figure 27 is the block diagram that explanation utilizes dynamic image encoding device 50 usefulness of embodiments of the present invention 5.

This dynamic image encoding device 50, the candidate screen of when P picture and B picture are encoded, carrying out the forward direction reference, and for different from the dynamic image encoding device 10 of above-mentioned execution mode 1 on the coded system of B picture.

Namely, this dynamic image encoding device 50, coding-control unit 110 and the mode selected cell 109 of replacement execution mode 1 have the coding-control unit 150 and the mode selected cell 159 that carry out the action different from them.

Specifically, the coding-control unit 150 of the dynamic image encoding device 50 of execution mode 5, to reference picture memory 117 control so that, when the P picture is encoded, 4 pictures (I or P picture) that are positioned at this P picture the place ahead are used as the candidate screen of forward direction reference, when the B picture is encoded, is positioned at 4 pictures (I or P picture) in this B picture the place ahead, 1 I that is positioned at immediate 1 the forward direction B picture of this B picture and rear or P picture and is used as candidate screen.

In addition, the mode selected cell 159 of execution mode 5, to the piece (object piece) of P picture when encoding, as the coded system of object piece, in picture coding, use the inter-picture prediction coding of motion vector and do not use the inter-picture prediction coding of motion vector (motion is made as 0 to be processed) and select one.To the piece (object piece) of B picture when encoding, as the coded system of object piece, in picture coding, use forward motion vector the inter-picture prediction coding, use the inter-picture prediction coding of backward motion vector and use the inter-picture prediction coding of forward motion vector and backward motion vector and select one.Namely, the mode selected cell 159 of the dynamic image encoding device 50 of this execution mode 5, mode selected cell 109 with the dynamic image encoding device 10 of execution mode 1, only do not using difference on the direct mode this point, therefore, this dynamic image encoding device 50 does not have the motion vector storage unit 116 of dynamic image encoding device 10.

In addition, the part beyond the coding-control unit 150 of the dynamic image encoding device 50 of this execution mode 5 and the mode selected cell 159, identical with in the dynamic image encoding device 10 of execution mode 1.

Then, the action of dynamic image encoding device 50 described.

The input picture is input to input picture memory 101 by the displaying time order with picture unit.Shown in Figure 29 (a), be input to input picture memory 101 by displaying time order each picture P0, B1, B2, P3, B4, B5, P6, B7, B8, P9, B10, B11, P12, B13, B14, P15, B16, B17 and P18.

Each picture of inputting in input picture memory 101 shown in Figure 29 (b), changes row by coded sequence.This changes the reference relation that row is based in the inter-picture prediction coding and carries out.Namely, carry out the row that changes of the picture of inputting, so that when the 1st picture is encoded, be used as the 2nd picture with reference to the candidate of picture, encode first than the 1st picture.

In this execution mode 5, when P picture (object picture) was encoded, 4 pictures (I or P picture) that approach that are in time its place ahead were used as the candidate with reference to picture.In addition, when the B picture coding, be in time 4 pictures (I or P picture) in its place ahead that approaches, be in immediate 1 B picture in its place ahead and be in the immediate I at rear or the P picture is used as candidate with reference to picture in time.

Carried out changing each picture of row in the picture memory 101 in input, be read out by the unit of each motion compensation.In this execution mode 5, the unit of motion compensation is 16 pixels of level * vertical 16 pixel sizes, namely Pixel arrangement is become rectangular rectangular area, the pixel count of horizontal direction and the pixel count of vertical direction are the zone (macro block) of 16 pixels.In the following description, macro block is referred to as piece.

Below, the coding of picture P15, B13, B14 processed describing in this order.

The coding of＜picture P15 is processed 〉

Because picture P15 is the P picture, so this picture is implemented to use the inter-picture prediction coding of forward direction reference.In addition, in the process that the P picture is encoded, the B picture is not used as with reference to picture.

Figure 28 represents the situation with reference to the management of the picture in the picture memory 117.

For example, in the coding zero hour of picture P15, in reference picture memory 117, on the storage area that has been endowed the logical storage numbering, the ascending order according to the logical storage numbering stores picture P12, B11, P9, P6, P3.These pictures are encoded, and the view data of storing in reference picture memory 117 is the view data of being decoded by dynamic image encoding device 50.Below, for the purpose of simplifying the description, will in memory, store the picture of its view data, also referred to as the picture of in memory, storing.

To in reference picture memory 117, store with reference to candidate screen, by the control of coding-control unit 150, carry out giving with reference to picture index.This index is given, and is not by the coded sequence of picture but the time sequencing when showing is carried out.Particularly, be exactly in chronological order to new with reference to candidate screen, the order that namely shows slow with reference to candidate screen, the value of giving less with reference to picture index.But, when the P picture coding, the B picture is not carried out giving with reference to picture index.In addition, when the B picture is encoded, in chronological order to up-to-date with reference to candidate screen, give expression as the backward symbol of processing with reference to picture [b].

According to so definite method with reference to picture index, as shown in figure 28, to picture P12, P9, P6, P3 give with reference to picture index, be respectively [0], [1], [2], [3], B11 does not give with reference to picture index to picture.

In addition, when the P picture coding, coding-control unit 150 is controlled so that switch 113,114,115 becomes connection each switch.Piece from the picture P15 that reads of input picture memory 101 is input to motion vector detecting unit 108, mode selected cell 159 and calculus of differences unit 102.

Motion vector detecting unit 108 will be in the picture that is stored in reference to picture memory 117, and picture P12, the P9, P6, the P3 that are endowed with reference to picture index use as the reference candidate screen, the piece among the picture P15 are carried out the detection of motion vector.In the case, select optimally with reference to candidate screen from picture P12, P9, P6 and P3, the detection of motion vector is carried out with reference to picture with reference to selected.Afterwards, with detected motion vector, to mode selected cell 159 and 104 outputs of bit string generation unit.In addition, motion vector is the information Rp with reference to which resulting motion vector of picture P12, P9, P6, P3 and so on, namely also exports to mode selected cell 159 with reference to picture index.

Mode selected cell 159 uses by the detected motion vector of motion vector detecting unit 108, determines the coded system of the piece of picture P15.Coded system, expression with which kind of method is encoded to piece.For example, to the piece of P picture, as coded system, select coding in the picture, use motion vector the inter-picture prediction coding, do not use in the inter-picture prediction coding of motion vector (motion is made as 0 to be processed).The selection of coded system generally is to make under predetermined figure place encoding error for minimum.

By mode selected cell 159 determined coded system Ms, output to bit string generation unit 104.In addition, be to carry out in the situation of coded system of forward direction reference in determined coded system, with reference to picture index also to 104 outputs of bit string generation unit.

In addition, the resulting predicted picture Pd of coded system based on being determined by mode selected cell 159 is output to calculus of differences unit 102 and add operation unit 106.But, under having selected the situation of encoding in the picture, not prediction of output image Pd.In addition, in the situation of having been selected to encode in the picture by mode selected cell 159, make input terminal Ta be connected to lead-out terminal Tb2 by switch 111 is controlled, make lead-out terminal Td be connected to input terminal Tc2 by switch 112 is controlled.

Below, select the situation of inter-picture prediction coding to describe to mode selected cell 159.Because the action of in the case calculus of differences unit 102, coded prediction error unit 103, bit string generation unit 104 and predicated error decoding unit 105 is identical with execution mode 1, so its explanation is omitted.

When the coding processing of whole piece of picture P15 finished, coding-control unit 150 carried out corresponding to the logical storage numbering of the picture of storing in reference picture memory 117 with reference to the renewal of picture index.

Namely, because the picture P15 that has encoded is newer than which picture of storing in reference picture memory 117 by the displaying time order, so picture P15 just stores the storage area that is set logical storage numbering (0) into.Then, the logical storage numbering with reference to the storage area of picture that has stored other just increases progressively 1.And then, because next coded object picture is the picture B13 as the B picture, so picture B11 is also carried out giving with reference to picture index.Thus, in the storage area that is set logical storage numbering (0)～(5), stored picture P15, P12, B11, P9, P6, P3 give respectively with reference to picture index [0], [1], [2], [3], [4] picture P12, B11, P9, P6, P3 respectively.Because becoming next object picture is the B picture, so to being stored in the picture P15 of logical storage numbering 0, replace with reference to picture index, gives expression and be used as the backward symbol of processing with reference to picture [b].

The coding of＜picture B13 is processed 〉

Because picture B13 is the B picture, so use the inter-picture prediction coding of two-way reference.In the case, as the candidate screen of forward direction reference, use 4 I or the P picture that approach in time, and immediate 1 B picture in time, as the candidate screen of backward reference, use to be in time immediate I or P picture.Thereby, be picture P12, B11, P9, P6, P3 just for the candidate screen of the forward direction reference of picture B13, and be picture P15 just for the candidate screen of the backward reference of picture B13.These are stored in reference in the picture memory 117 with reference to candidate screen.With reference to candidate screen, as shown in figure 28, give logical storage numbering and with reference to picture index to these.

When coding B picture, coding-control unit 150 is controlled so that switch 113,114,115 becomes connection each switch.Thereby the piece from the picture B13 that reads of input picture memory 101 is imported into motion vector detecting unit 108, mode selected cell 109, calculus of differences unit 102.

In motion vector detecting unit 108, the candidate screen that is used as the forward direction reference with reference to picture P12, B11 in the picture memory 117, P9, P6, P3 will be stored in, picture P15 as the candidate screen of backward reference, is carried out the forward motion vector of piece of picture B13 and the detection of backward motion vector.In the case, the detection of forward motion vector, by from picture P12, B11, P9, P6, P3 with optimal picture and selecting for to carry out with reference to picture.Detected motion vector is to mode selected cell 159 and 104 outputs of bit string generation unit.In addition, forward motion vector is which the information Rp of motion vector and so on reference to picture P12, B11, P9, P6, P3, namely with reference to picture index also to 159 outputs of mode selected cell.

Then, the action of mode selected cell 159, calculus of differences unit 102, bit string generation unit 104 and predicated error decoding unit 105, identical when encoding with picture P15.

When the processing of the whole pieces among the picture B13 finishes, by the control of coding-control unit 150, carry out corresponding to the logical storage numbering of the picture of in reference picture memory 117, storing with reference to the renewal of picture index.

Namely, because picture B13, by the displaying time order than the picture P15 that in reference picture memory 117, stores also front, also rear, so just storing into, picture B13 is set the storage area that logical storage is numbered (1) than the picture P12 that in reference picture memory 117, stores.In addition, owing to be not used as with reference to picture in the processing of the coding of picture B11 picture afterwards, so deleted.At this moment, expression is output to bit string generation unit 104 with the information that picture B11 deletes from the reference picture memory as control signal Cs1.Bit string is recorded and narrated and outputed to bit string generation unit 104 with this information as heading message.In addition, the logical storage of the storage area of corresponding picture P12 numbering just increases progressively 1.

Next object picture to be encoded is the picture B14 as the B picture.Thereby the picture of storing in the storage area with logical storage numbering (0) is used as backward with reference to picture, and the picture beyond it is given with reference to picture index.Thus, in the storage area corresponding to logical storage numbering (0)～(5), distinguish stored picture P15, B13, P12, P9, P6, P3, and picture B13, P12, P9, P6, P3 are given respectively with reference to picture index [0], [1], [2], [3], [4].

The coding of＜picture B14 is processed 〉

Because picture B14 is the B picture, so use the inter-picture prediction coding of two-way reference.In this case, as picture B14 with reference to picture, use picture B13, P12, P9, P6, P3 as forward direction with reference to picture, use picture P15 as backward with reference to picture.In the processing of B picture, coding-control unit 150 is controlled so that switch 113,114,115 becomes connection each switch.Thereby the piece of the picture B14 that reads from input picture memory 101 is imported into motion vector detecting unit 108, mode selected cell 159, calculus of differences unit 102.

Motion vector detecting unit 108, the candidate screen that is used as the forward direction reference with reference to picture B13, P12 in the picture memory 117, P9, P6, P3 will be stored in, picture P15 as the candidate screen of backward reference, is carried out the forward motion vector of piece of picture B14 and the detection of backward motion vector.In the case, the detection of forward motion vector is by becoming to carry out with reference to picture with optimal picture and selecting from picture B13, P12, P9, P6, P3.Detected motion vector is output to mode selected cell 159 and bit string generation unit 104.In addition, forward motion vector is the information Rp with reference to which resulting motion vector of picture B13, P12, P9, P6, P3 and so on, namely with reference to picture index, also is output to mode selected cell 159.

Then, the action of mode selected cell 159, calculus of differences unit 102, bit string generation unit 104, predicated error decoding unit 105 and add operation unit 106, substantially identical when being encoded with picture P15.

When the processing of the whole pieces among the picture B14 finishes, by the control of coding-control unit 150, carry out corresponding to the logical storage numbering of the picture of in reference picture memory 117, storing with reference to the renewal of picture index.

Namely, because picture B14, by the displaying time order than the picture P15 that in reference picture memory 117, stores also front, also rear, so just storing into, picture B14 is set the storage area that logical storage is numbered (1) than the picture B13 that in reference picture memory 117, stores.In addition, so owing to be not used as with reference to picture deleted in the processing of the coding of picture B13 picture afterwards.At this moment, expression is output to bit string generation unit 104 with the information that picture B13 deletes from the reference picture memory as control signal Cs1.Bit string generation unit 104 is recorded and narrated this information in bit string as heading message.

Next object picture to be encoded is the picture P18 as the P picture.Correspondingly, the offscreen picture of B is given with reference to picture index.Thus, in the storage area of logical storage numbering (0)～(5), stored picture P15, B14, P12, P9, P6 give respectively with reference to picture index [0], [1], [2], [3] picture P15, P12, P9, P6 respectively.

As mentioned above, in present embodiment 5, because in a plurality of candidate screen for the forward direction reference of coded object picture, the slower just value of giving of its displaying time less with reference to picture index (forward motion vector that namely is used for being identified in the detected object piece is the information of having used which candidate screen), so in candidate screen as with reference to the easiest selecteed picture of picture, just the value of giving less with reference to picture index.Thus, just can make the size of code minimum for the reference picture index, and can obtain the raising of code efficiency.

Below, to the effect of present embodiment 5, take the coding for the B picture be with other B picture as the situation of carrying out with reference to candidate screen as example, the problem points that simultaneously prior art is existed describes particularly.

For example, suppose that the picture of moving image is arranged as shown in Figure 29 (a), the forward direction when encoding for the object picture uses the situation of 4 P pictures and 1 B picture to describe with reference to candidate screen.

Figure 30 represents the management example with reference to the picture in the picture memory in the case.In addition, in this memory, candidate screen is stored by coded sequence.

When picture P15 encoded, the ascending order according to the logical storage numbering in the reference picture memory stored picture B11, P12, P9, P6, P3 as candidate screen.In addition, to these candidate screen, give respectively with reference to picture index [0], [1], [2, [3], [4].Thereby, the Shi Buhui that is encoded at the P picture as the B picture (here being picture B11) with reference to picture, has been given with reference to picture index.Therefore, produce can not be used with reference to picture index, can cause the deteriorated of code efficiency.

In addition, when picture B13 encodes, in the reference picture memory, store picture P15, B11, P12, P9, P6, P3 by the logical storage indexed sequential.Give expression as the backward symbol [b] that uses with reference to picture to picture P15, to remaining picture, give respectively with reference to picture index [0], [1], [2, [3], [4].Thereby the picture B13 that adjusts the distance (object picture) is picture B11 far away in time, and than apart from this picture B13 near picture P12 in time, it is little of picture index that the value of giving is also wanted.General in the situation of motion detection, since on the probability apart from the object picture in time near candidate screen be used as with reference to the situation of picture many, so when the value with reference to picture index of picture B11 far away in time become than near in time picture P12 with reference to the also little value of picture index the time, can cause the deteriorated of code efficiency.

And then, when picture B14 is encoded, in the reference picture memory, store picture B13, P15, B11, P12, P9, P6 by the logical storage indexed sequential.Give expression as the backward symbol [b] that uses with reference to picture to picture B13, to remaining picture, give respectively with reference to picture index [0], [1], [2, [3], [4].Thereby the former backward picture P15 with reference to picture that should be used as picture B14 but is used as the candidate screen of forward direction reference.And the picture B13 of the candidate of the former forward direction reference that should be used as picture B14 but is used as backward with reference to candidate screen.As a result, correctly encode with regard to difficult.In addition, when picture B14 encodes, then be present in reference in the picture memory as the picture B11 that is not used as with reference to the picture of picture.

On the other hand, in embodiments of the present invention 5, as shown in figure 28, because in the reference picture memory, press DISPLAY ORDER storage object picture with reference to candidate screen, in the candidate screen of forward direction reference, the slower just value of giving of its displaying time less with reference to picture index, so in candidate screen, as with reference to the easiest selecteed picture of picture, just give less of picture index.Thus, just can make the size of code minimum for the reference picture index, can obtain the raising of code efficiency.

In addition, when the P picture coding, because the B picture is not carried out with reference to the giving of picture index, so just can avoid the generation with reference to picture index that can not be used, further improve code efficiency.

In addition, when the B picture coding, owing to the picture in the storage area that the logical storage that is stored in the respective value minimum is numbered, do not give reference and use picture index, and this picture is used with reference to picture as backward.So, when the predictive coding of B picture, just can avoid answering the P picture of backward reference but to be used as forward direction with reference to the situation of picture.

In addition, when becoming the picture that is not used as with reference to picture from reference picture memory deletion, in bit string, record and narrate the information that deletion has been carried out in expression.So, just can detect in the decoding side and delete the picture that the decoding of decoder object picture and later picture thereof is processed, is not used as with reference to picture from the reference picture memory.

In addition, in above-mentioned execution mode 5, motion compensation is take the image space (macro block) that is comprised of 16 pixels of level * vertical 16 pixels as unit, and the coding of prediction error image is processed take the image space (sub-block) that is comprised of 8 pixels of level * vertical 8 pixels as unit.But as the macro block of the unit of motion compensation (coding of prediction error image) or the pixel count of sub-block, can be from above-mentioned different.

In addition, in above-mentioned execution mode 5, the number of continuous B picture is 2, but the number of above-mentioned continuous B picture also can be 3 or more.

In addition, in above-mentioned execution mode 5, be 4 to the quantity of the candidate screen of forward direction reference when the P picture is encoded, but it also can be other value beyond 4.

In addition, in above-mentioned execution mode 5, the candidate screen of forward direction reference is 4 P pictures and 1 B picture during to the B picture coding, but the candidate screen of the forward direction reference of B picture is not limited to these pictures.

In addition, in above-mentioned execution mode 5, become each picture coded object, that consist of moving image, when other later pictures of this each picture are encoded, be used as with reference to picture.But a plurality of pictures that consist of moving image also can comprise the picture that is not used as with reference to picture.In the case, by in the reference picture memory, not storing the picture that is not used as with reference to picture, just obtain the effect same with above-mentioned execution mode 5.

In addition, in above-mentioned execution mode 5, for the coding of B picture be with other B picture as carrying out with reference to candidate screen, but can be not do not carry out with reference to other B picture for the coding of B picture yet.In the case, in the reference picture memory, do not store the B picture.Even perhaps also sequentially give with reference to picture index according to the displaying time of picture under these circumstances, just can obtain the effect same with above-mentioned execution mode 5.

In addition, in above-mentioned execution mode 5, be give 1 system with reference to picture index, but also can be respectively to forward, backward give different system with reference to picture index.

In addition, in this execution mode 5, in the candidate screen of forward direction reference, the slower just value of giving of its displaying time less with reference to picture index, but if conduct is with reference to the easiest selecteed picture of picture in candidate screen, just the value of giving less with reference to picture index, then be not limited to above-mentioned explanation.

In addition, Figure 31 is expression corresponding to the concept map of having given with reference to the structure (form of coding image signal) of the bit string of the picture of picture index.

The code signal Pt that is equivalent to 1 picture includes the heading message Hp that comprises in the picture beginning, and the data portion Dp after this.In heading message Hp, include control signal (RPSL).The coded data (bit string) that in data portion Dp, includes in addition corresponding each piece.

For example, bit string BLx is the bit string with the piece of coded system coding in the picture, and bit string BLy is the bit string with the piece of the coding of the inter-picture prediction coded system beyond the coded system in the picture.

In the bit string BLx of piece, include heading message Hbx, about the information Prx of coded system and the image information Dbx that is encoded.In the bit string BLy of piece, include heading message Hby, about the information Pry of coded system, the 1st with reference to picture index Rld1, the 2nd with reference to picture index Rld2, the 1st motion vector MV1, the 2nd motion vector MV2 and the image information Dby that is encoded.Can judge by the information Pry relevant with coded system be use the 1st, the 2nd with reference among picture index Rld1, the Rld2 which.

At first, forward direction is preferentially given with reference to picture index Rld1 with reference to candidate screen with reference to candidate screen is relatively backward.Make and backwardly preferentially give with reference to picture index Rld2 with reference to candidate screen with reference to the relative forward direction of candidate screen.

[execution mode 6]

Figure 32 is the block diagram that explanation utilizes moving image decoding apparatus 60 usefulness of embodiments of the present invention 6.

The moving image decoding apparatus 60 of this execution mode 6 is decoded to the bit string Bs that exports from the dynamic image encoding device 50 of above-mentioned execution mode 5.

The moving image decoding apparatus 60 of this execution mode 6, the candidate screen of when P picture and B picture are decoded, carrying out the forward direction reference, and the coded system of B picture is different from the moving image decoding apparatus 20 of above-mentioned execution mode 2.

Namely, the moving image decoding apparatus 60 of this execution mode 6, memory control unit 204 and the mode decoding unit 223 of replacement execution mode 2 have the memory control unit 264 and the mode decoding unit 263 that carry out the action different from them.

Specifically, the memory control unit 264 of the moving image decoding apparatus 60 of execution mode 6, to reference picture memory 207 control so that, when the P picture is decoded, 4 pictures (I or P picture) that are positioned at this P picture the place ahead are used as the candidate screen of forward direction reference, when the B picture is decoded, is positioned at 4 pictures (I or P picture) in this B picture the place ahead, 1 I that is positioned at immediate 1 the forward direction B picture of this B picture and rear or P picture and is used as candidate screen.

In addition, the mode decoding unit 263 of the moving image decoding apparatus 60 of execution mode 6, when piece (object piece) decoding of P picture, as the coded system of object piece, decoding in picture, use the inter-picture prediction decoding of motion vector and do not use the inter-picture prediction decoding of motion vector (will move and process as 0) one of selection.When piece (object piece) decoding of B picture, as the decoding process of object piece, decoding in picture, use forward motion vector the inter-picture prediction decoding, use the inter-picture prediction decoding of backward motion vector and use the inter-picture prediction decoding of forward motion vector and backward motion vector and select one.

Namely, the mode decoding unit 263 of the moving image decoding apparatus 60 of this execution mode 6, mode decoding unit 223 with the moving image decoding apparatus 20 of execution mode 2, only process on this point different in the decoding of not using corresponding direct mode, thereby, this moving image decoding apparatus 60 does not have the motion vector storage unit 226 of moving image decoding apparatus 20.

In addition, the moving image decoding apparatus 60 of execution mode 6 is except memory control unit 264 and mode decoding unit 263, identical with the moving image decoding apparatus 20 of execution mode 2.

Then, the action of dynamic image encoding device 60 described.

Input is from the bit string Bs of dynamic image encoding device 50 outputs of execution mode 5 in moving image decoding apparatus shown in Figure 32 60.In this bit string Bs, be to be in time 4 I nearby in its place ahead or P picture as carry out the inter-picture prediction coding with reference to candidate screen for each P picture.In addition, each B picture is that 4 P pictures, 1 the forward direction B picture that will be in time its immediate the place ahead and the immediate I that is in time its rear or P picture are as encoding with reference to candidate screen.

In the case, the order of the picture in the bit string is shown in Figure 29 (b).

Below, the decoding of picture P15, B13, B14 processed describing in this order.

The decoding of＜picture P15 is processed 〉

The bit string of picture P15 is imported into bit string analytic unit 201.Bit string analytic unit 201 extracts various data from the input bit string.Various data are the information of coded system and motion vector etc.The information that the mode that extracts is selected (coded system) Ms is output to mode decoding unit 263.In addition, the motion vector MV that extracts is output to motion compensation decoding unit 205.And then, coded prediction error data Ed is exported to predicated error decoding unit 202.

Mode decoding unit 263, the coded system Ms with reference to extracting from bit string carries out the control of switch 209 and 210.In the situation of in coded system is picture, encoding, switch 209 is controlled so that input terminal Te is connected to lead-out terminal Tf1, switch 210 is controlled so that lead-out terminal Th is connected to input terminal Tg1.On the other hand, be in the situation of inter-picture prediction coding in coded system, switch 209 is controlled so that input terminal Te is connected to lead-out terminal Tf2, switch 210 is controlled so that lead-out terminal Th is connected to input terminal Tg2.

In addition, mode decoding unit 263 also outputs to motion compensation decoding unit 205 with coded system Ms.

Below, be that the situation that inter-picture prediction is encoded describes to coded system.

Predicated error decoding unit 202 is decoded generation forecast error information PDd to the coded data Ed that inputs.The prediction error data PDd that generates outputs to switch 209.Because the input terminal Te of switch 209 is connected to lead-out terminal Tf1, so prediction error data PDd outputs to add operation unit 208.

Motion compensation decoding unit 205 according to the information of the motion vector of inputting etc., generates motion compensated image.Be input to the information of motion compensation decoding unit 205, for motion vector MV with reference to picture index Rp.Motion compensated image (predicted picture) based on these information, is obtained from reference picture memory 207 in motion compensation decoding unit 205.Picture P15 is that picture P12, P9, P6, P3 are encoded as the candidate of reference picture, and these pictures are decoded and be stored in reference to picture memory 207.

Figure 28 is illustrated in the picture of storing with reference in the picture memory 207.As shown in figure 28, when picture P15 is decoded, in reference picture memory 207, store picture P12, B11, P9, P6, P3.

Memory control unit 264 carries out giving with reference to picture index to what store with reference to candidate screen in reference picture memory 207.This gives with reference to picture index, based on the displaying time of picture sequentially carry out so that up-to-date picture value of being endowed little with reference to picture index.When the P picture is decoded, the B picture is not carried out giving with reference to picture index.Thereby, corresponding to picture P12, P9, P6, P3 be respectively [0], [1], [2], [3] with reference to picture index, and picture B11 is not given with reference to picture index.

Motion compensation decoding unit 205 according to the reference picture index, determines when the object piece is encoded it is which picture with reference to picture P12, P9, P6, P3.Then, motion compensation decoding unit 205,, is obtained predicted picture (prediction data Pd) and is generated motion compensated image (predicted picture) from reference picture memory 207 with reference to picture and motion vector based on determined.The motion compensated image that generates like this is output to add operation unit 208.

Add operation unit 208 generates decoded picture (data Ad) with prediction error image and the motion compensated image addition of inputting.The decoded picture that generates outputs to reference to picture memory 207 by switch 210.

When whole macro blocks of picture P15 were decoded, memory control unit 264 carried out corresponding to the logical storage of the picture of storing in reference picture memory 207 numbering with reference to the renewal of picture index.

At this moment, because picture P15 is newer than any picture of storing in reference picture memory 207 in chronological order, so picture P15 just stores the storage area that is set logical storage numbering (0) into.Then, corresponding stored there is the logical storage numbering of storage area of other candidate screen increase progressively 1.

In addition, because next decoder object picture is picture B13, so picture B11 is carried out giving with reference to picture index.Thus, in the storage area that is set logical storage numbering (0)～(5), stored picture P15, P12, B11, P9, P6, P3 give respectively with reference to picture index [0], [1], [2], [3], [4] picture P12, B11, P9, P6, P3 respectively.

The decoding of＜picture B13 is processed 〉

Because the action of bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, identical when processing with the decoding of picture P15 is so their action specification is omitted.

Motion compensation decoding unit 205 is from the Information generation moving compensating data of motion vector of inputting etc.The information that is input to motion compensation decoding unit 205 is motion vector and with reference to picture index.Picture B11 encodes with picture P15 the candidate screen that picture P12, B11, P9, P6, P3 are used as the forward direction reference as the candidate screen of backward reference.In the decoding of picture B13 constantly, these candidate screen are decoded and be stored in reference to picture memory 207.

Be that motion compensation decoding unit 205 according to the reference picture index, is determined when picture B13 is encoded, and carries out the forward direction reference with which candidate screen of picture P12, B11, P9, P6, P3 in forward prediction or the bi-directional predicted situation in coded system.Then, motion compensation decoding unit 205 is obtained forward motion compensation image with reference to picture and forward motion vector from reference picture memory 207 based on determined.Be that motion compensation decoding unit 205 is obtained reverse compensating images with reference to picture and backward motion vector from reference picture memory 207 based on determined in the situation of bi-directional predicted predictive pictures coding or back forecast coding in coded system.Then, motion compensation decoding unit 205 uses forward motion compensation image and reverse compensating images, generates motion compensated image (predictive picture).

The motion compensated image that generates like this is output to add operation unit 208.Add operation unit 208 generates decoded picture with prediction error image and the motion compensated image addition of inputting.The decoded picture that generates outputs to reference to picture memory 207 by switch 210.

When whole pieces of picture B13 were decoded, memory control unit 264 carried out corresponding to the logical storage of the picture of storing in reference picture memory 207 numbering with reference to the renewal of picture index.Because picture B13, also more forward than the picture P15 that in reference picture memory 207, stores by the displaying time order, after also will leaning on than the picture P12 that in reference picture memory 207, stores, so picture B13 just stores the storage area that is set logical storage numbering (1) into.

In addition, because in bit string, recording and narrating has expression with the information of picture B11 from the deletion of reference picture memory, so 264 pairs of memory control units are controlled so that picture B11 is deleted from reference picture memory 207 with reference to picture memory.

In addition, the logical storage numbering with reference to the storage area of candidate screen P12 that stores other is increased progressively 1.Thus, in the storage area that is set logical storage numbering (0)～(5), stored picture P15, B13, P12, P9, P6, P3 give respectively with reference to picture index [0], [1], [2], [3], [4] picture B13, P12, P9, P6, P3 respectively.

The decoding of＜picture B14 is processed 〉

Because the action of bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, identical when processing with the decoding of picture P15 is so their action specification is omitted.

Motion compensation decoding unit 205 is from the Information generation moving compensating data of motion vector of inputting etc.The information that is input to motion compensation decoding unit 205 is motion vector and with reference to picture index.Picture B14 is the candidate screen that picture B13, P12, P9, P6, P3 is used as the forward direction reference, picture P15 is encoded as the candidate screen of backward reference, in the decoding of picture B14 constantly, these candidate screen are decoded and be stored in reference to picture memory 207.

When coded system was forward predictive coded or bi-directional predictive coding, motion compensation decoding unit 205, determined when picture B14 is encoded which picture of picture B13, P12, P9, P6, P3 is by the forward direction reference according to the reference picture index.Then, motion compensation decoding unit 205 is obtained forward motion compensation image with reference to picture and forward motion vector from reference picture memory 207 based on determined.Be that motion compensation decoding unit 205 is obtained reverse compensating images with reference to picture and backward motion vector from reference picture memory 207 based on determined in the situation of predictive pictures coding of bi-directional predicted or back forecast in coded system.Then, motion compensation decoding unit 205 uses forward motion compensation image and reverse compensating images, generates motion compensated image.

The motion compensated image that generates like this outputs to add operation unit 208.By add operation unit 208, prediction error image and the motion compensated image addition of inputting generated decoded picture.The decoded picture that generates outputs to reference to picture memory 207 by switch 210.

When the processing of the whole pieces among the picture B14 finishes, by memory control unit 264, carry out corresponding to the logical storage numbering of the picture of in reference picture memory 207, storing with reference to the renewal of picture index.Because picture B14, also more forward than the picture P15 that in reference picture memory 207, stores by the displaying time order, after also leaning on than the picture B13 that in reference picture memory 207, stores, so picture B14 just stores the storage area that is set logical storage numbering (1) into.In addition, because in bit string, recording and narrating has expression with the information of picture B13 from the deletion of reference picture memory, picture B13 is from 207 deletions of reference picture memory so 264 pairs of these memories of memory control unit are controlled.

Because next decoder object picture is the picture P18 as the P picture, and the offscreen picture of B is given with reference to picture index.Thus, in the storage area that is set logical storage numbering (0)～(5), distinguish stored picture P15, B14, P12, P9, P6, P3, and picture P15, P12, P9, P6 are given respectively with reference to picture index [0], [1], [2], [3] and [4].

In addition, from reference picture memory 207, by the displaying time order picture of decoding is exported as output image.

Below, each picture thereafter is by decoding with corresponding similarly processing of picture type.

As mentioned above, in present embodiment 6, in a plurality of candidate screen corresponding to the forward direction reference of decoder object picture, the slower just value of giving of its displaying time less with reference to picture index (forward motion vector that namely is used for the identifying object piece is with reference to the resulting information of which candidate screen), and based in the bit string of object picture, comprise with reference to picture index, from above-mentioned a plurality of candidate screen, determine with reference to picture.So, in candidate screen as with reference to the easiest selecteed picture of picture, just the value of giving less with reference to picture index.Thus, just can make the size of code minimum with reference to picture index, and can be to correctly being decoded by the resulting bit string of the high coding method of code efficiency.

In addition, when the P picture codec, because the B picture is not carried out with reference to the giving of picture index, thus just can avoid the generation with reference to picture index that can not be used, and can be to correctly being decoded by the resulting bit string of the high coding method of code efficiency.

In addition, because when the B picture codec, the picture that in the storage area of the minimum logical storage numbering of the value of being set, store, process with reference to picture as backward, this picture is not carried out giving with reference to picture index, so when the predictive coding of B picture, just can avoid the P picture to be used as forward direction with reference to the situation of picture, and can be to correctly being decoded by the resulting bit string of the high coding method of code efficiency.

In addition, recorded and narrated in the situation of bit string becoming the information that is not used as with reference to the picture of picture from reference picture memory deletion, based on this information from the deletion of reference picture memory with reference to picture, so just can effectively utilize with reference to picture memory.

In addition, in above-mentioned execution mode 6, the arrangement as a plurality of pictures that consist of moving image illustrates the arrangement of the picture of 2 B pictures between the P of adjacency picture.But the number of the B picture between the P of adjacency picture also can be other value, for example, and 3 or 4.

In addition, in above-mentioned execution mode 6, be 4 for the quantity of the candidate screen of the forward direction reference of P picture, but it also can be other value.

In above-mentioned execution mode 6, for the candidate screen of the forward direction reference of B picture, be 4 P pictures and 1 B picture, but as the candidate screen for the forward direction reference of B picture, be not limited to this.

In addition, in above-mentioned execution mode 6, each picture that consists of moving image is used as with reference to picture when other later pictures of this each picture are decoded, but a plurality of pictures of moving image also can comprise the picture that is not used as with reference to picture.In the case, by in the reference picture memory, not storing the picture that is not used as with reference to picture, just can obtain the effect same with above-mentioned execution mode 6.

In addition, in above-mentioned execution mode 6, the decoding of B picture be with other B picture as carrying out with reference to candidate screen, but can be not do not carry out with reference to other B picture for the decoding of B picture yet.In the case, in the reference picture memory, do not store the B picture.Also be in the case, sequentially give with reference to picture index according to the displaying time of picture, just can obtain the effect same with above-mentioned execution mode 6.

In above-mentioned execution mode 6, for the purpose of simplifying the description, to being used for management with reference to the memory of candidate screen, decoded picture is not illustrated with reference to picture memory as 1 by the memory that DISPLAY ORDER changes row and output with making any distinction between with being used for, but moving image decoding apparatus 60, also can have respectively for management with reference to the diode-capacitor storage of candidate screen, and be used for decoded picture and change row's memory by what DISPLAY ORDER changed row and output.

In the case, diode-capacitor storage is controlled by memory control unit 264, outputs to motion compensation decoding unit 205 with reference to candidate screen.In addition, change the decoded picture image that row's memory will be arranged by decoding order, change row and successively output by DISPLAY ORDER.

In addition, in above-mentioned execution mode 6, show by 1 rule and give situation with reference to picture index to candidate screen, be 1 system with reference to picture index namely, but as with reference to picture index, also can be as illustrated in the enforcement mode 5, use 2 systems with reference to picture index.

[execution mode 7]

Figure 33 is the block diagram that explanation utilizes dynamic image encoding device 70 usefulness of embodiments of the present invention 7.

This dynamic image encoding device 70, the candidate screen of when P picture and B picture are encoded, carrying out the forward direction reference, and for different from the dynamic image encoding device 10 of above-mentioned execution mode 1 on the coded system of B picture.

Namely, this dynamic image encoding device 70, coding-control unit 110 and the mode selected cell 109 of replacement execution mode 1 have the coding-control unit 170 and the mode selected cell 179 that carry out the action different from them.

Specifically, the coding-control unit 170 of execution mode 7, to reference picture memory 117 control so that, when the P picture coding, 3 pictures (I or P picture) that are positioned at this P picture the place ahead are used as the candidate screen of forward direction reference, and when the B picture coding, be positioned at 2 pictures (I or P picture) in this B picture the place ahead, 1 I that is positioned at immediate 1 the forward direction B picture of this B picture and rear or P picture and be used as candidate screen.But, be positioned at the B picture of relative object picture in more the place ahead of the I in immediate the place ahead or P picture and be not referenced.

Above-mentioned coding-control unit 170 is controlled so that the sign whether the indicated object picture is referenced when processing for the coding of the later picture of this object picture is inserted in the bit string corresponding to the object picture according to control signal Cd bit strings generation unit 104.Specifically, be exactly that bit string generation unit 104 is controlled by control signal Cd so that be illustrated in decoding and the time data of object picture should be stored into reference to the information in the picture memory 117, and represent that it should store during information append in the bit string.

In addition, the mode selected cell 179 of execution mode 7, when piece (object piece) coding of P picture, coded system as the object piece, in picture, encode, use the inter-picture prediction coding of motion vector, do not use in the inter-picture prediction coding of motion vector (motion is made as 0 to be processed) and select one, when the piece coding of B picture, as the coded system of object piece, in picture, encode, use the inter-picture prediction coding of forward motion vector, use the inter-picture prediction coding of backward motion vector, and select one in the inter-picture prediction of use forward motion vector and the backward motion vector coding.Namely, the mode selected cell 179 of the dynamic image encoding device 70 of this execution mode 7, mode selected cell 109 with the dynamic image encoding device 10 of execution mode 1, only do not using difference on the direct mode this point, thereby, this dynamic image encoding device 70 does not have the motion vector storage unit 116 of dynamic image encoding device 10.Other assemblies of the dynamic image encoding device 70 of execution mode 7, all with the dynamic image encoding device 10 of execution mode 1 in identical.

The dynamic image encoding device 70 of this execution mode 7, with the dynamic image encoding device 50 of execution mode 5,170 bit strings generation units 104 are controlled so that the sign whether the indicated object picture is referenced when processing for the coding of the later picture of this object picture is inserted in the bit string different on this point in the coding-control unit.Particularly, utilize control signal Cd that coding generation unit 104 is controlled, so that the sign whether the indicated object picture is referenced when processing for the coding of the later picture of this object picture is inserted in the bit string corresponding to the object picture.In addition, dynamic image encoding device 70, also different from dynamic image encoding device 50 on P picture and B picture are encoded the candidate screen of time institute's reference.Except above-mentioned difference, other each points in this dynamic image encoding device 70 are all identical with dynamic image encoding device 50.

Then, the action of dynamic image encoding device 70 described.

Input image data Id is input to input picture memory 101 with picture unit in chronological order.

Figure 34 (a) is illustrated in the order of the picture of inputting in the input picture memory 101.

In input picture memory 101, shown in Figure 34 (a), each picture begins to input in order from picture P1.In Figure 34 (a), picture P1, P4, P7, P10, P13, P16, P19, P22 are the P pictures, and picture B2, B3, B5, B6, B8, B9, B11, B12, B14, B15, B17, B18, B20, B21 are the B pictures.

When the P picture coding, this P picture 3 pictures (I or P picture) nearby of being in time the place ahead are used as the candidate with reference to picture relatively.In addition, when the B picture was encoded, this B picture was in time the immediate I at 2 pictures (I or P picture), immediate 1 the B picture that is in the place ahead and rear in its place ahead nearby or P picture and is used as candidate with reference to picture relatively.But when the B picture coding, the B picture that is located at more the place ahead of the I in immediate the place ahead or P picture is not referenced.In addition, when the I picture is encoded, not with reference to other picture.

The data I d of each picture of inputting in input picture memory 101 changes row by coded sequence.Below with the data of each picture just referred to as picture.

In other words, the order of picture is changed the row that changes who lines up coded sequence from input sequence process, be based on the reference relation of inter-picture prediction coding in processing and carry out.Change during row processes at this, each picture is changed row so that be used as the 2nd picture with reference to the candidate of picture when the 1st picture coding, also formerly encode than the 1st picture.

Specifically, from each picture of picture P10～P13 with reference to the corresponding relation between the candidate screen, shown in the arrow among Figure 34 (a).Namely, when P picture P10 was encoded, picture P1, P4, P7 were referenced, and when P picture P13 was encoded, picture P4, P7, P10 were referenced.In addition, when B picture B11 was encoded, picture P7, P10, P13 were referenced, and when B picture B12 was encoded, picture P7, P10, B11, P13 were referenced.

Figure 34 (b) expression has carried out changing the order of the picture behind the row to the picture from the picture B2 shown in Figure 34 (a) to P22.After changing row, each picture is arranged by the order of P4, B2, B3, P7, B5, B6, P10, B8, B9, P13, B11, B12, P16, B14, B15, P19, B17, B18, P22.

In input picture memory 101, carried out changing each picture of row, to each predetermined data processing unit, be read out according to the scramble time order.In this execution mode 7, data processing unit is the data cell of carrying out motion compensation, more specifically, is a rectangular image space (macro block) of all having arranged in the horizontal direction 16 pixels with vertical direction.In the following description, with macro block referred to as piece.

Below, the coding of picture P13, B11, B12 processed describing in order.

The coding of＜picture P13 is processed 〉

Because picture P13 is the P picture, encode so process the inter-picture prediction of using the forward direction reference as the coding of picture P13.In the case, be 3 I or the P picture that relative object picture is positioned at the place ahead with reference to candidate screen, specifically be exactly picture P4, P7, P10.These are with reference to candidate screen, and coding is processed and finished, and are stored in reference to picture memory 117 corresponding to the data Dd of decoded picture.

In the process that the P picture is encoded, coding-control unit 170 is controlled so that switch 113,114,115 becomes connection each switch.

The data M d corresponding with the piece of the picture P13 that reads from input picture memory 101 is input to motion vector detecting unit 108, mode selected cell 179, calculus of differences unit 102.

Motion vector detecting unit 108 is utilized to be stored in reference to picture P4, P7 in the picture memory 117, the decode image data Rd of P10, carries out the detection of motion vector MV of the piece of picture P13.In the case, select optimally with reference to picture from picture P4, P7, P10, and the detection of motion vector is by carrying out with reference to selected picture.Afterwards, detected motion vector MV is output to mode selected cell 179 and bit string generation unit 104.In addition, expression with reference among picture P4, P7, the P10 which when detecting motion vector MV, be subject to reference to and so on information (with reference to image information), also be output to mode selected cell 179.

In mode selected cell 179, use by the detected motion vector of motion vector detecting unit 108, determine the coded system of the piece of picture P13.

Specifically, be exactly in the situation of P picture coding, the inter-picture prediction of coding, use motion vector is encoded, is not used the inter-picture prediction coding of motion vector (motion is made as 0 to be processed) in picture, determines with which method to encode.In the determining of coded system, generally be to select to make to give as size of code in the position of quantity that will be predetermined that encoding error is minimum method in the situation of piece.

By mode selected cell 179 determined coded system Ms, be output to bit string generation unit 104.In addition, be to carry out in the situation of coded system of forward direction reference at determined coded system Ms, when being illustrated in forward motion vector and detecting with reference to which the information (with reference to image information) of picture P4, P7, P10, also to 104 outputs of bit string generation unit.

Afterwards, from mode selected cell 179 determined coded system Ms accordingly with reference to the resulting predicted image data Pd of picture, be output to calculus of differences unit 102 and add operation unit 106.But, under having selected the situation of encoding in the picture, do not export above-mentioned predicted image data Pd.In addition, selected identical in switch 111 and 112 control and the execution mode 5 in the situation of coding in the picture.

Below, just by mode selected cell 179 as coding mode selection the situation of inter-picture prediction coding describe.

The action of calculus of differences unit 102, coded prediction error unit 103, bit string generation unit 104 and predicated error decoding unit 105 and coding-control unit 170, identical with in the execution mode 5.

But, in this execution mode 7, as the heading message of picture P13, additional representation picture P13 encodes 3 I or the P picture in the place ahead as the reference candidate screen information.In addition, because picture P13 will be referenced when other pictures are encoded, so the decoded data Dd of corresponding picture P13 should be stored into reference to the information in the picture memory 117 (sign) when being illustrated in decoding, also be added as the heading message of picture P13.And then, expression picture P13 should be stored into reference to during in the picture memory 117 until finish the decoded information of picture P22, also add as the heading message of picture P13.

Method as the above-mentioned memory period that represents picture 13, the temporal information of with good grounds picture P22 (for example, time-based positional information, such as picture index, decode time information or displaying time information etc.) method that represents, method that represents according to information (such as frame numbers etc.) during from picture P13 to picture P22 etc.Above-mentioned heading message can be used as the heading message take picture as unit, namely as the heading message to each object picture, records and narrates.Also can be used as the heading message of sequence integral body or take frame as unit the heading message of (for example GOP unit in the MPEG mode) record and narrate.

In the coded system for each piece of picture P13, in the situation for the coded system of carrying out the forward direction reference, being illustrated in when detecting forward motion vector is which information (with reference to image information) with reference among picture P4, P7, the P10 also is affixed in the bit string.For example, in the situation that motion vector obtains for reference picture P10, the P picture in tight the place ahead of presentation code object picture just appends in the bit string for the information (with reference to picture index) with reference to picture.In the situation that motion vector obtains for reference picture P7, the P picture before 2 of presentation code object picture just appends in the bit string for the information (with reference to picture index) with reference to picture.Motion vector be with reference to the resulting situation of picture P4 under, the P picture before 3 of presentation code object picture just appends in the bit string for the information (with reference to picture index) with reference to picture.For example, can be with reference to picture by the picture in tight the place ahead of reference picture index [0] indicated object picture, being with reference to picture by the picture before 2 of reference picture index [1] indicated object picture, is with reference to picture by the picture before 3 of reference picture index [2] indicated object picture.

In addition, as heading message, record and narrate expression P picture and use 3 information of carrying out the inter-picture prediction coding with reference to candidate screen.

By processing same as described above, carry out the coding processing for the remaining macro block of picture P13.When the coding processing for whole macro block of picture P13 finishes, then carry out the coding of picture B11 and process.

The coding of＜picture B11 is processed 〉

Because picture B11 is the B picture, encode so process the inter-picture prediction of using two-way reference as the coding of picture B11.In the case, use 2 pictures (I or P picture) of approaching in time apart from object picture (picture B11) and immediate 1 B picture in time as the candidate screen of forward direction reference, as the candidate screen use of backward reference apart from coded object picture in time immediate I or P picture.But not with reference to its immediate I that surpasses the object picture or the B picture of P picture.

Thereby, use picture P7, P10 as the forward direction of picture B11 with reference to picture, use picture P13 as the backward of picture B11 with reference to picture.In the processing of the 1st B picture of 2 continuous B pictures, because this B picture of the 1st is used as with reference to picture when another B picture is encoded, so coding-control unit 170 is controlled so that switch 113,114,115 becomes connection each switch.Therefore, the view data Md corresponding with the piece of the picture B11 that reads from input picture memory 101 is imported into motion vector detecting unit 108, mode selected cell 179, calculus of differences unit 102.

Motion vector detecting unit 108, the picture P7, the P10 that are stored in reference to picture memory 117 are used as the candidate screen of forward direction reference, the picture P13 that is stored in reference to picture memory 117 is used with reference to picture as backward, carry out the detection of forward motion vector and the backward motion vector corresponding with the piece of picture B11.In the case, the detection of forward motion vector is carried out with reference to picture by selection is optimal from picture P7, P10.Detected motion vector is to mode selected cell 179 and 104 outputs of bit string generation unit.In addition, forward motion vector is the information (with reference to image information) of which detected motion vector with reference to picture P7, P10 and so on, also to 179 outputs of mode selected cell.

Mode selected cell 179 uses by the detected motion vector of motion vector detecting unit 108, determines the coded system corresponding with the piece of picture B11.As the coded system of B picture, for example in the picture coded system, with the inter-picture prediction coded system of forward motion vector, select with the inter-picture prediction coded system of backward motion vector and with the inter-picture prediction coded system of bi-directional motion vector.In the selection of this coded system, also as described above, select to make quantity in the position that will be scheduled to give as size of code that encoding error is minimum conventional method (mode) in the situation of piece.

By mode selected cell 179 determined coded systems, be output to bit string generation unit 104.In addition, from mode selected cell 179 determined coded system Ms accordingly with reference to the resulting predicted image data Pd of picture, be output to calculus of differences unit 102 and add operation unit 106.But, under having selected the situation of encoding in the picture, not prediction of output view data Pd.In addition, in the situation of having been selected to encode in the picture by mode selected cell 179, to switch 111 and 112, same control when carrying out with above-mentioned picture P13 coding.

Below, the situation of having been selected the inter-picture prediction coding by mode selected cell 179 is described.

In the case, the action of calculus of differences unit 102, coded prediction error unit 103, bit string generation unit 104 and predicated error decoding unit 105 and coding-control unit 170, identical with in the execution mode 5.

Be to carry out in the situation of coded system of forward direction reference in coded system, reference was which information (with reference to image information) among picture P7, the P10 also is affixed in the bit string when expression was detected forward motion vector.For example, in reference is in the situation of picture P10, be exactly presentation code object picture tight the place ahead candidate screen for reference to picture with reference to image information, in the situation of reference picture P7, be exactly before 2 of presentation code object picture candidate screen for reference to picture with reference to image information.For example, can be with reference to picture by the candidate screen in tight the place ahead of reference picture index [0] indicated object picture, be with reference to picture by the candidate screen in tight the place ahead of reference picture index [1] indicated object picture.

In addition, in the case, the B picture of presentation code object be with the B picture in the place ahead also as the information of implementing the inter-picture prediction coding with reference to picture and processing, do not advance additional as heading message.In addition, the place ahead of the B picture of presentation code object be the information of 2 I or P picture and 1 B picture with reference to candidate screen, add as heading message.And then expression with reference to the information of the B picture that more is positioned at the place ahead than I or the P picture in the immediate the place ahead of relative coding object B picture, is not advanced additional as heading message.

Thus, when the bit string Bs that the dynamic image encoding device 70 by present embodiment 7 is generated decodes, just can know need to have which kind of memory capacity with reference to picture memory.Above-mentioned these heading messages can be used as the heading message take picture as unit, namely as the heading message to each object picture, record and narrate.Also can be used as the heading message of sequence integral body or take several pictures as unit the heading message of (such as GOP in the MPEG mode etc.) record and narrate.

In addition, because picture B11 is used as with reference to picture to the picture coding at its rear the time, so when decoding, the decode image data Dd corresponding to this picture B11 should be stored into the information with reference to picture memory 117 and so on, and be illustrated in reference to should store in the picture memory 117 these data Dd during for until the decoding of the picture B12 information when finishing add as heading message respectively.

When remaining coding processing for picture B11 finishes, then carry out the coding of picture B12 and process.

The coding of＜picture B12 is processed 〉

Because picture B12 is the B picture, encode so process the inter-picture prediction of using two-way reference as the coding of picture B12.In the case, use 2 I or the P picture that approaches in time apart from coded object picture B12 as the candidate screen of forward direction reference, apart from this coded object picture B12 immediate B picture in time.In addition, the candidate screen use as backward reference is in immediate I or P picture in time apart from this object picture B12.Specifically, be picture P7, P10, B11 for the candidate screen of the forward direction reference of picture B12, be picture P13 for the backward of picture B12 with reference to picture.

Because picture B12 is not used as with reference to picture to other picture codings the time, so coding-control unit 170 according to control signal Cs1 each switch is controlled so that switch 113 becomes connection, and switch 114,115 becomes disconnection.Therefore, the view data Md corresponding with the piece of the picture B12 that reads from input picture memory 101 is imported into motion vector detecting unit 108, mode selected cell 179, calculus of differences unit 102.

Motion vector detecting unit 108, picture P7, the P10, the B11 that are stored in reference to picture memory 117 are used as forward direction with reference to picture, with being stored in reference to the picture P13 of picture memory 117 as backward with reference to picture, carry out the detection of forward motion vector and the backward motion vector corresponding with the piece of picture B12.

In the case, the detection of forward motion vector is carried out with reference to picture by selection is optimal from picture P7, P10, B11.Detected motion vector is output to mode selected cell 179 and bit string generation unit 104.In addition, forward motion vector is the information (with reference to image information) of which detected motion vector with reference to picture P7, P10, B11 and so on, also to 179 outputs of mode selected cell.

Mode selected cell 179 uses by the detected motion vector of motion vector detecting unit 108, determines to be applied to the coded system of the piece of picture B12.As the coded system of B picture, in the picture coded system, with the inter-picture prediction coded system of forward motion vector, with selecting among the inter-picture prediction coded system of backward motion vector and the inter-picture prediction coded system with bi-directional motion vector.

By mode selected cell 179 determined coded system Ms, to 104 outputs of bit string generation unit.In addition, from mode selected cell 179 determined coded systems accordingly with reference to the resulting predicted image data Pd of picture, be output to calculus of differences unit 102 and add operation unit 106.But, in the situation of in selecting picture, encoding, not prediction of output view data Pd.

In addition, in the situation of being selected to encode in the picture by mode selected cell 179, switch 111 and 112, the control mode when being encoded with above-mentioned picture P13 is identical.

Below, the situation of having been selected the inter-picture prediction coding by mode selected cell 179 is described.

In the case, the action of calculus of differences unit 102, coded prediction error unit 103, bit string generation unit 104 and predicated error decoding unit 105 and coding-control unit 170, identical with in the execution mode 5.

Be to carry out in the situation of coded system of forward direction reference in coded system, forward motion vector also is affixed in the bit string for the information (with reference to image information) with reference to which detected motion vector of picture P7, P10, B11 and so on.

In addition, as heading message, presentation code object B picture B12 also implements the information that the inter-picture prediction coding is processed as the candidate of reference picture with the B picture B11 in its place ahead, and the candidate screen of expression forward direction reference is that the information of 2 I or P picture and 1 B picture is is also recorded and narrated.

In addition, as heading message, added represent that picture B12 is not used as with reference to the information of picture when the coding of the picture at its rear is processed.

Thus, just can judge easily that the decode image data Dd of corresponding picture B12 does not need to store into reference to picture memory when decoding, thereby make the management with reference to picture memory become simple and easy.

Above-mentioned these heading messages can be recorded and narrated with picture unit as heading message, namely record and narrate as the heading message of each object picture to be encoded.In addition, also can be used as the heading message of sequence integral body or record and narrate with the heading message of several picture unit (such as GOP in the MPEG mode etc.).

By same processing, carry out the remaining coding of picture B12 and process.

Then, the view data corresponding with later each picture of above-mentioned picture B12 by corresponding with picture type, encoded with above-mentioned same processing.For example, P picture and picture P13 similarly process, and the 1st B picture (picture B14, B17 etc.) of continuous B picture is processed equally with picture B11.The 2nd of continuous B picture the B picture (picture B15, B18 etc.) is similarly processed with picture B12 in addition.

Like this, in the dynamic image encoding device 70 of present embodiment 7, when as the object picture B picture being encoded, because the candidate screen as the forward direction reference, beyond the P picture, also use the B picture, so with reference to picture, just can utilize the immediate forward direction that is positioned at the coded object picture with reference to picture as forward direction.Thus, just can improve the precision of prediction for the motion compensation of B picture, can obtain the raising of code efficiency.

And when as the object picture B picture being encoded, as heading message, whether this object picture of additional representation (during decoding) when other picture codings is used as the information with reference to picture.And then, as heading message, additional representation this object picture during at other picture codings (during decoding) be used as this coded object picture to be stored into reference to the information during the picture memory in the situation with reference to the information of picture.So when the bit string Bs from 70 outputs of this dynamic image encoding device is decoded, just can recognize at an easy rate and when decoding, in picture memory, store what picture, perhaps this memory period how long, thereby the management with reference to picture memory in the time of can making decoding is simplified.

In this execution mode 7, the object B picture with other B picture as the situation about processing of implementing with reference to picture to encode under, this situation is recorded and narrated as the heading message of the B picture of coded object.But this heading message might not be recorded and narrated with picture unit.Also can be used as the heading message of sequence integral body or record and narrate with the heading message of several picture unit (such as GOP in the MPEG mode etc.).

In present embodiment 7, motion compensation is to carry out take the macro block that is comprised of 16 pixels (horizontal direction) * 16 pixels (vertical direction) as unit, it is take the piece that is comprised of 4 pixels (horizontal direction) * 4 pixels (vertical direction) as unit that the coding of prediction error image data is processed, and perhaps processes take the piece that is comprised of 8 pixels (horizontal direction) * 8 pixels (vertical direction) as unit.But the coding of motion compensation and prediction error image data is processed, and the image space that also can form take the pixel of the number different from these units carries out as unit.

In addition, in present embodiment 7, the coded system of P picture be in picture coded system, use the inter-picture prediction coded system of motion vector and do not use the inter-picture prediction coded system of motion vector and select 1 coded system, the coded system of B picture be in picture coded system, use forward motion vector the inter-picture prediction coded system, use backward motion vector the inter-picture prediction coded system, use the inter-picture prediction coded system of bi-directional motion vector and select 1 coded system.But select the method for the coded system of P picture and B picture, be not limited to above-mentioned execution mode 7.

In addition, in present embodiment 7, adopt between I picture and P picture, and the image sequence that between the P of adjacency picture, inserts 2 B pictures.But between I picture and P picture, and the number of the B picture that inserts between the P picture of adjacency is not 2 yet in image sequence, such as, be 3 or 4.

In addition, although in present embodiment 7, be 3 to the frame numbers that is used as the candidate screen of forward direction reference when the P picture coding, it also can be other value.

In addition, although in present embodiment 7, as the candidate screen of forward direction reference, use 2 P pictures and 1 B picture when the B picture coding, the candidate screen of the forward direction reference of conduct when the code of B picture also can be not limited to this.For example, as the candidate screen for the forward direction reference of B picture, can use 1 P picture and 2 B pictures, 2 P pictures and 2 B pictures or with picture type it doesn't matter ground in time the position near 3 pictures of object picture.

When the B picture coding, only with the position near 1 picture of this B picture as with reference to picture, the expression piece is just not need to record and narrate in bit string with reference to the information (with reference to image information) which picture is encoded and so on.

In addition, in present embodiment 7, when the B picture coding, the B picture than the position is positioned at the place ahead more near forward direction I or the P picture of this B picture is not used as with reference to picture.But when the B picture coding, also can will more be positioned at the B picture in the place ahead than the immediate forward direction P picture that is positioned at this B picture, be used as with reference to picture.

[execution mode 8]

Figure 35 is the block diagram that explanation utilizes moving image decoding apparatus 80 usefulness of embodiments of the present invention 8.

The moving image decoding apparatus 80 of this execution mode 8 is decoded to the bit string Bs that exports from the dynamic image encoding device 70 of above-mentioned execution mode 7.

This moving image decoding apparatus 80 carries out on the candidate screen of forward direction reference when P picture and B picture codec, and for different from the moving image decoding apparatus 20 of above-mentioned execution mode 2 on the decoding process of B picture.

Namely, this moving image decoding apparatus 80, memory control unit 204 and the mode decoding unit 223 of replacement execution mode 2 have the memory control unit 284 and the mode decoding unit 283 that carry out the action different from them.

Specifically, the memory control unit 284 of execution mode 8, reference picture memory 287 is controlled so that when the P picture codec, 3 pictures (I or P picture) that are positioned at this P picture the place ahead are used as the candidate screen of forward direction reference, when the B picture codec, be positioned at 2 pictures (I or P picture) in this B picture the place ahead, 1 I that is positioned at immediate 1 the forward direction B picture of this B picture and rear or P picture and be used as candidate screen.But be set in relative object picture and do not carry out reference at the I in its immediate the place ahead or the B picture more on the front of P picture.

Above-mentioned memory control unit 284, based in the bit string corresponding to the object picture, insert, represent whether this object picture carries out reference when processing for the coding of the later picture of this object picture sign, utilize control signal Cm to reference to controlling with picture memory 207.

Specifically, in the bit string corresponding to the object picture, comprise and be illustrated in when decoding and the data of object picture should be stored into information (sign) with reference to picture memory 207, and represent that it should store during information.

In addition, the mode decoding unit 283 of execution mode 8, when the piece decoding of P picture, as the decoding process of object piece, decoding in the picture, select one among using the inter-picture prediction decoding of motion vector and not using the inter-picture prediction decoding of motion vector (motion is made as 0 to be processed).When the piece decoding of B picture, as the decoding process of object piece, decoding in the picture, use the inter-picture prediction decoding of forward motion vector, select one among using the inter-picture prediction decoding of backward motion vector and using the inter-picture prediction decoding of forward motion vector and backward motion vector.Namely, the mode decoding unit 283 of the moving image decoding apparatus 80 of this execution mode 8, mode decoding unit 223 with the moving image decoding apparatus 20 of execution mode 2, only do not using difference on the direct mode this point, thereby, this moving image decoding apparatus 80 does not have the motion vector storage unit 226 of moving image decoding apparatus 20.And, other assemblies of the moving image decoding apparatus 80 of execution mode 8, identical with in the moving image decoding apparatus 20 of execution mode 2.

In addition, the moving image decoding apparatus 80 of this execution mode 8, with the moving image decoding apparatus 60 of execution mode 6, control to be illustrated in whether this object picture is referenced when processing for the coding of the later picture of object picture sign different on this point in being inserted into corresponding to the bit string of this object picture at memory control unit 284 bit strings generation units 104.In addition, in moving image decoding apparatus 80, also different from the moving image decoding apparatus 60 of above-mentioned execution mode 6 in the candidate screen of P picture and B picture codec time institute reference.Other assemblies in the moving image decoding apparatus 80 of this execution mode 8, identical with the moving image decoding apparatus 60 of execution mode 6.

Then, 80 actions describe to dynamic image encoding device.

In the moving image decoding apparatus 80 of this execution mode 8, the bit string Bs that input is generated by the dynamic image encoding device 70 of above-mentioned execution mode 7.

In this execution mode 8, when the P picture codec, 3 pictures (I or P picture) that approach that relative this P picture are in time the place ahead use as the candidate with reference to picture.On the other hand, when the B picture codec, 1 I or the P picture that relative this B picture are in time 2 pictures (I or P picture) in approaching the place ahead, immediate 1 the B picture that is in the place ahead and rear use as the candidate of reference picture.But, when the B picture is decoded, do not carry out reference to being positioned at the I that is in time immediate the place ahead or the B picture more on the front of P picture.In addition, when the I picture codec, not with reference to other picture.

In addition, which picture is used as the heading message Ih that information with reference to picture is used as bit string Bs and records and narrates when being illustrated in P picture or B picture codec, in candidate screen, and this heading message, is extracted by bit string analytic unit 201.

This heading message Ih is output to memory control unit 284.This heading message can be used as the heading message of sequence integral body, the heading message, the heading message of picture unit of (such as GOP in the MPEF mode etc.) are recorded and narrated take several pictures as unit.

Be input to the order of the picture among the bit string Bs of this moving image decoding apparatus 80, shown in Figure 36 (a), be the decoding order of picture.Below, the decoding of picture P13, B11, B12 processed describing particularly in order.

The decoding of＜picture P13 is processed 〉

When the bit string of corresponding picture P13 was imported into bit string analytic unit 201, bit string analytic unit 201 extracted various data from the bit string of input.Various data are to select the information of relevant information (coded system) Ms, motion vector MV and above-mentioned heading message Ih etc. with mode.The coded system Ms that extracts is output to mode decoding unit 283.In addition, the motion vector MV that extracts is output to motion compensation decoding unit 205.And then the coded data Ed that is extracted by bit string analytic unit 201 is output to predicated error decoding unit 202.

Mode decoding unit 283, information (coded system) Ms with reference to the mode that extracts from bit string is selected controls switch 209 and 210.Coded system in coded system Ms is picture, and when coded system Ms is the inter-picture prediction coded system to 209 and 210 control, identical with above-mentioned execution mode 6.

In addition, mode decoding unit 283 outputs to motion compensation decoding unit 205 with coded system Ms.

Below, be that the situation of inter-picture prediction coded system describes to selection mode.

Because the action of predicated error decoding unit 202, motion compensation decoding unit 205, add operation unit 208, identical with in the execution mode 6 is so its explanation is omitted.

Figure 37 represents stored by reference picture memory 207 the time dependent situation of picture of data.

When the decoding of picture P13 begins, in regional R1, the R2 of reference picture memory 207, R3, store picture B8, P7, P10.To picture P13, picture P7, P10 are processed as implement decoding with reference to the candidate of picture, and in the storage area R1 that has stored picture B8 stored picture P13.The rewriting with reference to the view data of each picture in the picture memory like this is based on that the heading message of each additional in bit string picture carries out.This heading message is to be illustrated in reference in the picture memory 207, until handling Chengdu, the decoding of picture P13 answers stored picture P7, until handling Chengdu, the decoding of picture P16 answers stored picture P10, until the information that stored picture B8 is answered in Chengdu is handled in the decoding of picture B9.

In other words, no longer need in the decoded process of the later picture of picture P13 owing to can judge picture B8, thus stored picture B8 with reference to picture storage area R1, P13 is rewritten by picture.

In addition, because as the heading message of picture P13, recording and narrating and represent that picture P13 finishes up to the decoding of picture P19 and should be stored in reference to the information in the picture memory, so be stored in reference in the picture memory till during at least to this.

As mentioned above, the piece among the picture P13 is decoded in order.When whole when decoded corresponding to the coded data of the piece of picture P13, just begin to process for the decoding of picture B11.

The decoding of＜picture B11 is processed 〉

Because identical with during the decoding of picture P13 is processed of the action of bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, so to their action specification omission.

Motion compensation decoding unit 205 according to the information of inputting, such as the information of motion vector etc., generates motion-compensated image data Pd.Namely, the information that is input to motion compensation decoding unit 205 be with corresponding to the motion vector MV of above-mentioned picture B11 with reference to the relevant information of picture index.Picture B11 carries out reference as forward direction with reference to picture with picture P10, and picture P13 is carried out reference with reference to picture and implements coding and process as backward.Thereby when picture B11 decoded, these candidate screen P10, P13 were through decoding, and corresponding decode image data DId is stored in reference to picture memory 207.

Be in the situation of bi-directional predicted predictive pictures coded system in coded system, motion compensation decoding unit 205, information based on the expression forward motion vector, obtain forward direction with reference to image from reference picture memory 207, information based on the expression backward motion vector obtains backward with reference to image from reference picture memory 207.Then, motion compensation decoding unit 205 by to forward direction with reference to image with backwardly carry out summation averaging with reference to image, generate motion compensated image.The data Pd of the motion compensated image that generates like this is output to add operation unit 208.

Add operation unit 208, with prediction error image data PDd and the motion-compensated image data Pd addition of inputting, output addition view data Ad.The addition view data Ad that so generates outputs to reference to picture memory 207 by switch 210 as decode image data DId.

With reference to picture memory control unit 284, based on as the heading message of bit string, expression P picture and B picture are which candidate screen to implement the information of encoding and processing with reference to, carry out the control with reference to picture memory 207.

Figure 37 represents with reference to the time dependent situation of the store status of the picture in the picture memory 207.

When the decoding of picture B11 begins, in reference picture memory 207, store picture P13, P7, P10.Picture B11 processes stored picture B11 in the storage area R2 that has stored picture P7 with picture P10, P13 as implement decoding with reference to picture.The rewriting with reference to the view data of each picture in the picture memory like this is based on that the heading message of each additional in bit string picture carries out.This heading message is to be illustrated in reference in the picture memory 207, Shi Douying stores picture P7 until the decoding of picture P13 is finished dealing with, Shi Douying stores picture P10 until the decoding of picture P16 is finished dealing with, and Shi Douying carries out canned data to picture P13 until the decoding of picture P19 is finished dealing with.

In other words owing to can judge that picture P7 no longer needs when the later picture of picture P13 is decoded, so just stored picture P7 with reference to picture storage area R2 in, stored picture B11.

In addition, because the heading message as picture B11, expression picture B11 is arranged in record until Shi Douying is finished in the decoding of picture B12 is stored in reference to the information in the picture memory 207, so till during at least to this, picture B11 is stored in reference in the picture memory 207.

As mentioned above, decode in order corresponding to the coded data of the piece of picture B11.When whole when decoded corresponding to the coded data of the piece of picture B11, carry out the decoding of picture B12.

The decoding of＜picture B12 is processed 〉

Because the action of above-mentioned bit string analytic unit 201, mode decoding unit 203 and predicated error decoding unit 202, the action when processing with picture P13 decoding is identical, so their action specification is omitted.

Motion compensation decoding unit 205 according to the information of the motion vector of inputting etc., generates motion-compensated image data Pd.Namely, the information that is input to motion compensation decoding unit 205 is corresponding to the motion vector of picture B12 with reference to picture index.Picture B12 uses as forward direction picture P10 and B11 with reference to the candidate of picture, picture P13 is carried out reference with reference to picture and implements the coding processing as backward.These candidate screen P10, B11, P13 are through decoding, and corresponding decode image data is stored in reference to picture memory 207.

Be in the situation of bi-directional predictive coding mode in coded system, motion compensation decoding unit 205, determine that according to the reference picture index picture B12 implements coding as forward direction with reference to picture with which picture of picture P10, B11 and processes, then, information based on the expression forward motion vector obtains forward direction with reference to image from reference picture memory 207.In addition, motion compensation decoding unit 205, the information based on the expression backward motion vector obtains backward with reference to image from reference picture memory 207.Then, by to forward direction with reference to image with backwardly carry out summation averaging with reference to image, generate motion compensated image.The data Pd of the motion compensated image that so generates is 208 outputs to the add operation unit.

Add operation unit 208, with prediction error image data PDd and the motion-compensated image data Pd addition of input, output addition view data Ad.The addition view data Ad that so generates outputs to reference to picture memory 207 by switch 210 as decode image data DId.

Memory control unit 284 extracts based on the heading message from bit string, P picture and B picture be with which with reference to the information that picture is implemented to encode and processed and so on, carry out the control with reference to picture memory 207.

Figure 37 represents to go up in time the situation that changes with reference to the store status of the picture in the picture memory 207.When beginning is processed in the decoding of picture B12, in reference picture memory 207, store picture P13, B11, P10.Picture B12 processes picture P13, B11, P10 as implement decoding with reference to candidate screen.Because expression picture B12 is not used as information with reference to picture and so on and is used as heading message and records and narrates when other pictures are decoded, so the decoded data of picture B12 just is not stored in reference in the picture memory 207, and export as output image data Od.

As mentioned above, decode in order corresponding to the coded data of the piece of picture B12.Be stored in the decode image data with reference to each picture in the picture memory 207, and be not stored in reference to the decode image data in the picture memory, shown in Figure 36 (b), press the displaying time order and implement to reset processing, Od exports as output image data.

Thereafter, each with the corresponding coded data of each picture according to picture type, by with above-mentioned same processing, implement decoding and process.

Particularly, P picture and picture P13 are similarly implemented processing that coded data is decoded, the 1st the B picture (picture B14, B17 etc.) of continuous B picture implemented to process with the same decoding of picture B11.In addition, the 2nd the B picture (picture B15, B18 etc.) of continuous B picture implemented the decoding processing same with picture B12.

As mentioned above, in the moving image decoding apparatus 80 of present embodiment 8, because when decoding B picture, the B picture is used as with reference to candidate screen, so, just can be when coding B picture, to by the candidate screen as the forward direction reference, except the P picture, also use the coding of B picture to process resulting bit string and correctly decode.In addition and since use from bit string obtain represent P picture and B picture be with which with reference to picture implement the to encode information of processing and so on, carry out the control with reference to picture memory, so just become possibility with reference to effective utilization of picture memory.Namely, when being processed, afterwards decoding is used as the view data with reference to the picture of picture, remain on reference in the picture memory, and be not used as view data with reference to the picture of picture when deletion decoding is afterwards processed in order, just can effectively utilize with reference to picture memory.

Present embodiment 8 use corresponding to the bit string of inserting the image sequence of 2 B pictures between the P of adjacency picture, but the number of the B picture between the P of adjacency picture is not limited to 2, for example also can be 3 or 4.

In addition, in present embodiment 8, when the decoding of P picture is processed, use 2 pictures as the candidate screen of forward direction reference, but the quantity of carrying out the forward direction candidate screen of reference when the decoding of P picture is processed also can be the quantity beyond 2.

In addition, in present embodiment 8, when the B picture codec, candidate screen as the forward direction reference, to use 1 P picture and 1 B picture, and than relative decoder object B picture in time the immediate I in position or the P picture B picture that more is positioned at the place ahead be not used as with reference to picture, but the picture that when the decoding of B picture is processed, uses as the reference candidate screen, also can be from shown in the above-mentioned execution mode 8 different.In addition, when the decoding of B picture was processed, the immediate I in position or the P picture B picture that more is positioned at the place ahead also can be as with reference to picture in time than relative decoder object B picture.

In addition, in present embodiment 8, when other picture codecs, be not used as the decode image data with reference to the picture of picture, be not stored in reference to picture memory, but decode image data can be stored in this memory.

For example, when the output of the decode image data that makes each picture slightly postpones than the decoding of each picture, just the decode image data of each picture need to be stored into reference to picture memory.In the case, on the reference picture memory, the zone different from the storage area of storing as the decode image data of reference candidate screen is set, and in this zone, stores the decode image data that is not used as with reference to the picture of picture.In the case, although increase with reference to the memory capacity of picture memory, identical with reference to the management method of picture memory and above-mentioned execution mode 8, so can be easy to realize management with reference to picture memory.

Although in above-mentioned execution mode 2,4,6,8, all picture all is as with reference to candidate screen, not necessarily wants whole pictures all to be used as with reference to candidate screen.

In brief, usually, moving image decoding apparatus is no matter decoded picture is to be used as or to be not used as with reference to candidate screen, in case be stored in decoding buffer storage (decoded frame memory) afterwards, this decoded picture just read successively and shown from this decoding buffer storage.

In embodiments of the present invention 2,4,6,8, all pictures are used as with reference to candidate screen, thereby, decoded picture all be stored in storage as with reference to candidate screen with reference to picture memory, afterwards, decoded picture is read successively and is shown from the reference picture memory.

But, as mentioned above, be not decoded picture all be used as with reference to candidate screen.Thereby, also can make decoded picture, not only preserve the picture that is not used as with reference to candidate screen in case be stored in, but also preserve as with reference to the decoding buffer storage (decoded frame memory) of the picture of candidate screen afterwards, this decoded picture just read successively and shown from this decoding buffer storage.

In the respective embodiments described above, illustrate by hardware and realize dynamic image encoding device or moving image decoding apparatus, but these devices also can be realized by software.In the case, by will be used to the data storage medium that carries out the coding shown in the respective embodiments described above is processed or decoding is processed program and be recorded to floppy disc etc., independently just realizing at an easy rate above-mentioned dynamic image encoding device or moving image decoding apparatus in the computer system in advance.

Figure 38 (a)-38 (c) be for explanation carry out execution mode 1,3,5,7 any one moving image encoding is processed and the figure of the computer system that execution mode 2,4,6,8 any one moving picture decoding are processed.

Figure 38 (a) expression is included in front view, its cutaway view and the floppy disc main body D of floppy disc FD of the storage medium of the program of using in the computer system.The example of the physical format of Figure 38 (b) expression floppy disc main body D.

Floppy disc FD is built in above-mentioned disk main body D in the shell F, on the surface of this disk main body D, inwardly week forms a plurality of magnetic track Tr with concentric circles from periphery, and each magnetic track is split up into 16 sector Se along angle direction.Thereby, in having stored the floppy disc FD of said procedure, be used for carrying out the program that above-mentioned moving image encoding is processed or moving picture decoding is processed and just be recorded in the storage area (sector) of on above-mentioned disk main body D, giving.

In addition, Figure 38 (c) expression is used for carrying out the record of said procedure or the structure of reproduction at floppy disc FD.When said procedure being recorded among the floppy disc FD, from computer system Csys said procedure is written to floppy disc FD by floppy disk driver FDD.When the program in utilization is recorded in floppy disc FD is constructed above-mentioned dynamic image encoding device or moving image decoding apparatus in computer system, by floppy disk driver FDD computer system Csys is read and be sent to said procedure from floppy disc FD.

Although in the above description, be to give an example as an example of floppy disc example as recording medium, use CD also can similarly utilize the moving image encoding of software to process or the moving picture decoding processing with the situation of using above-mentioned floppy disc.In addition, recording medium is not limited to this, also can be CD-ROM, storage card, ROM cassette tape etc., as long as can logging program.In the situation of using these data carriers, also can similarly carry out moving image encoding processing or moving picture decoding processing by computer system with the situation of using above-mentioned floppy disc.

Below, to the application example of the dynamic image encoding method shown in the above-mentioned execution mode and dynamic image decoding method and use its system to describe.

Figure 39 is the integrally-built block diagram that expression realizes the content provider system 1100 of content converting service.

The district that provides of communication service is split up into the zone (residential quarter) of desired size, is respectively arranged with the base station 1107～1110 as fixed radio station in each residential quarter.

In this content provider system 1100, for example, by ISP 1102, telephone network 1104 and base station 1107～1110, computer 1111, PDA (personal digital assistant) 1112, camera 1113, portable phone 1114, be connected to internet 1101 with each equipment such as portable phone 1200 of camera.

But content provider system 1100 does not limit and will comprise all a plurality of equipment shown in Figure 39, can comprise the part of a plurality of equipment shown in Figure 39 yet.In addition, each equipment can not be directly connected to telephone network 1104 by the base station 1107～1110 as fixed radio station yet.

Camera 1113 be similar digital camera can be to the equipment of object movement image photography.In addition, portable phone is the portable telephone of PDC (Personal Digital Communication) mode, CDMA (Code Division Multiple Access) mode or GSM (Global System for Mobile Communications) mode, perhaps also can be PHS (Personal Handyphone System) etc., any one mode.

Streaming server 1103, by base station 1109, telephone network 1104 is connected with camera 1113.In this system, use camera 1113 can carry out the on-site delivery based on the data of the processing of encoding of user's transmission.The coding of captured data is both processed can be by camera 1113, also can be undertaken by the server of the transmission processing of carrying out data etc.Take resulting motion image data by 1116 pairs of moving images of camera and also can send to streaming server 1103 by computer 1111.Camera 1116 be wait can shot object still frame or the equipment of moving image, such as digital camera.In the case, which of the coding camera 1116 of motion image data, computer 1111 carries out all can.In addition, coding is processed the LSI1117 had by computer 1111 and camera 1116 and is carried out.

The software that encoding/decoding image is used can be stored in the storage medium (CD-ROM, floppy disc, hard disc etc.) as the recording medium that can be read by computer 1111 grades.And then motion image data also can be sent by the portable phone 1200 with camera.This motion image data is the data of being encoded and being processed by the LSI that portable phone 1200 has.

In this content provider system 1100, the content that the user takes by camera 1113 or camera 1116 etc. (for example, take the live video of concert etc.), similarly encoding with above-mentioned execution mode sends to streaming server 1103 from camera after processing.From streaming server 1103, the client computer of filing a request banishd the foregoing data send.

As client computer, the computer 1111, PDA1112, camera 1113, portable phone 1114 of the decoding data that can process above-mentioned coding etc. are arranged.

In such content provider system 1100, can receive and reproduce through coded data at client-side, and then by receiving the decode in real time at client-side, and reproduce, just can also realize private broadcast.

In the coding of each equipment that consists of this system, decoding, use dynamic image encoding device or the moving image decoding apparatus shown in the respective embodiments described above.

Example as moving image encoding or decoding device describes portable phone.

Figure 40 is the figure that the portable phone 1200 of dynamic image encoding method illustrated in above-mentioned any one execution mode and dynamic image decoding method is used in expression.

This portable phone 1200 has, be used for and base station 1110 between the antenna 1201 of transmitting-receiving electric wave, but the video image of shot object or the camera unit of still frame 1203, such as the CCD camera, show the display unit 1202 of liquid crystal display etc. of the data of the video image that receives by the captured video image of camera unit 1203, by antenna 1201 etc.

Portable phone 1200 has, the main unit 1204 of a plurality of operation keyss is installed, be used for carrying out the voice-output unit 1208 of the loud speaker etc. of voice output, be used for carrying out the voice-input unit 1205 of the microphone etc. of phonetic entry, the recording medium 1207 of the data that are used for keeping data, the data of moving image or the data of still frame etc. of the data of captured moving image or still frame, the mail that receives, coded data or decode is used for recording medium 1207 to be installed to the receptacle unit 1206 of portable phone 1200.

Recording medium 1207 has a kind of flash memory element of EEPROM (Electrically Erasable Programmable Read OnlyMemory) that can erasable nonvolatile memory as electricity in plastic casing, such as the SD card.

The below uses Figure 41 that portable phone 1200 is described in more detail.

Portable phone 1200 has the main control unit 1241 that each parts of the main unit that disposes display unit 1202 and operation keys 1204 are controlled blanketly.

Portable phone 1200 also has power circuit unit 1240, operation Input Control Element 1234, Image Coding unit 1242, camera interface unit 1233, LCD (LiquidCrystal Display) control unit 1232, image decoding unit 1239, demultiplexing/resolving cell 1238, recoding/reproduction unit 1237, modulation-demodulation circuit unit 1236 and Audio Processing Unit 1235.Each parts of portable phone 1200 interconnect by synchronous bus 1250.

Power circuit unit 1240, when becoming calling terminal/Power supply key for on-state by user operation, by by battery pack to each parts power supply, but the portable phone 1200 that starts with camera is operate condition.

In portable phone 1200, control each parts action by the control of the main control unit 1241 that formed by CPU, ROM and RAM etc.In specific words, in portable phone 1200, when the voice call mode, by to the resulting voice signal of the phonetic entry of voice-input unit 1205, be transformed into digital voice data by Audio Processing Unit 1235.Digital voice data is implemented spread processing by modulation-demodulation circuit unit 1236, and then implements digitaltoanalogconversion by transmission circuit unit 1231 and process and frequency conversion process, sends by antenna 1201.

In this portable phone 1200, the reception signal that is received by antenna 1201 when the voice call mode is exaggerated and implements frequency conversion process and analog-to-digital conversion is processed.Receive signal and implement contrary spread processing by modulation-demodulation circuit 1236 again, be transformed into analog voice signal by Audio Processing Unit 1235, this signal is by voice-output unit 1208 outputs.

Send during at data communication mode in the situation of Email at portable phone 1200, the text data of the Email that the operation of the operation keys 1204 by main unit is inputted sends to main control unit 1241 by operation Input Control Element 1234.Main control unit 1241, each parts is controlled so that by 1236 pairs of text datas enforcement of modulation-demodulation circuit spread processing, pass through antenna 1,201 1110 transmissions to the base station after implementing digitaltoanalogconversions processing and frequency conversion process by transmission circuit unit 1231.

When portable phone 1200 sends view data under data communication mode, supply to Image Coding unit 1242 by camera unit 1203 captured view data by camera interface unit 1233.In portable phone 1200, in the situation that does not send view data, also can will take resulting view data by camera unit 1203 by camera interface unit 1233 and LCD control unit 1232 and directly be shown in display unit 1202.

Image Coding unit 1242 disposes in the respective embodiments described above illustrated dynamic image encoding device in any one.This Image Coding unit 1242 is transformed into coded image data by by the dynamic image encoding method of any one in the above-mentioned execution mode view data of supplying with from camera unit 1203 being carried out compressed encoding, sends to demultiplexing/resolving cell 1238.In addition, simultaneously, portable phone 1200 will be input to voice-input unit 1205 by camera unit 1203 in photography voice send to demultiplexing/resolving cell 1238 by Audio Processing Unit 1235 as the speech data of numeral.

Demultiplexing/resolving cell 1238 carries out demultiplexing to the coded image datas of 1242 supplies from the Image Coding unit and the speech data of supplying with from Audio Processing Unit 1235 in a predetermined manner.The resulting demultiplexing data of its result are implemented spread processing by modulation-demodulation circuit 1236, and then implement digitaltoanalogconversion by transmission circuit unit 1231 and process and frequency conversion process, send by antenna 1201.

Portable phone 1200, when under data communication mode, being received in the data of the motion pictures files that links in homepage etc., by antenna 1201 from the base station the 1110 reception signals that receive, implement contrary spread processing by modulation-demodulation circuit 1236, the resulting demultiplexing data of its result send to demultiplexing/resolving cell 1238.

To the demultiplexing decoding data that receives by antenna 1201 time, demultiplexing/resolving cell 1238, by the demultiplexing data being decomposed the coding bit string that is divided into view data and the coding bit string of speech data, by synchronous bus 1250 this coded image data is supplied with image decoding unit 1239, simultaneously this speech data is supplied with Audio Processing Unit 1235.

Image decoding unit 1239 disposes and utilizes any one moving image decoding apparatus in the embodiments of the present invention.Image decoding unit 1239, by with the coding/decoding method corresponding to any one coding method in the above-mentioned execution mode, coding bit string to view data decodes to generate the reproducing motion pictures data, and by LCD control unit 1223 it is supplied with display unit 1202.Thus, carry out the demonstration of the motion image data that in the motion pictures files that for example homepage links, comprises.Meanwhile, audio treatment unit 1235 is supplied with voice-output unit 1208 with it after voice data is transformed into simulated audio signal.Thus, carry out the reproduction of the voice data that in the motion pictures files that for example homepage links, comprises.

, can use any one dynamic image encoding method in the above-mentioned execution mode and the system of dynamic image decoding method herein, and in the example that is limited to the foregoing feed system.

Recently, utilize the digital broadcasting of satellite or surface wave to become hot issue, the picture coding device of above-mentioned execution mode and picture decoding apparatus also can be applied to the digital broadcasting system as shown in figure 42 like that.

More specifically, gang up radio communication corresponding to the bits of coded of video information and 1409 sent to the satellites 1410 such as communication satellite or broadcasting satellite from the broadcasting station.When broadcasting satellite 1410 receives the coding bit string of above-mentioned video image information, the electric wave of output broadcasting usefulness, this electric wave is received by the antenna 1406 of the family that has the satellite broadcasting receiving equipment.For example, in the device of television set (receiver) 1401 or set-top box (STB) 1407 etc., the coding bit string is decoded, and reproducing video information.

In addition, read be recorded in as the coding bit string of the storage medium 1402 of the CD of recording medium and DVD etc. and the transcriber of decoding on, the moving image decoding apparatus shown in any one in the above-mentioned execution mode also can be installed.

In the case, the vision signal of reproduction is shown on the monitor 1404.Also can be at the set-top box 1407 interior mounted movable picture decoding apparatus of the antenna 1406 that is connected to cable 1405 that cable TV uses or the broadcasting of satellite/terrestrial ripple, reproduced by the output of 1408 pairs of these moving image decoding apparatus of monitor of television set.In the case, moving image decoding apparatus also can not be contained in set-top box, but be contained in the television set.In addition, also can be in the automobile 1412 with antenna 1411, receive signal from satellite 1410 or base station 1107 etc., on the display unit such as the auto navigation 1413 in being equipped on automobile 1412 moving image is reproduced.

And then, also can be encoded to picture signal by the dynamic image encoding device shown in any one in the above-mentioned execution mode, and be recorded to recording medium.

Tape deck at object lesson is register 1420, as with the DVD register of picture signal record to DVD dish 1421, and with picture signal record to the magnetic disk recorder of hard disc etc.Picture signal also can be recorded to SD card 1422.In addition, if register 1420 possesses the moving image decoding apparatus shown in any one in the above-mentioned execution mode, then can pass through register 1420, the picture signal that records in DVD dish 1421 and SD card 1422 is reproduced, be shown by monitor 1408.

Herein, structure as auto navigation 1413, can be included in the part beyond camera unit 1203 in the structure of portable phone shown in Figure 41, camera interface unit 1233, the Image Coding unit 1242, to computer 1111 and television set (receiver) 1401 etc. too.

In addition, as the terminal of portable phone 1114, can install: have both receiving type terminals of encoder/decoder, only have the transmitting terminal of encoder and only have 3 kinds of the receiving terminals etc. of decoder.

As mentioned above, in above-mentioned any equipment/system, the dynamic image encoding method shown in any one or dynamic image decoding method in the above-mentioned execution mode can be used, just illustrated in the above-described embodiment effect can be obtained by such processing.

In addition, need not superfluous words, embodiments of the present invention and application example thereof are not limited to shown in this specification.

As mentioned above, dynamic image encoding method and dynamic image decoding method according to the present invention, when the object picture that becomes coding or decoder object is the B picture, as for the object picture with reference to picture, can use the immediate forward direction picture that is positioned at this object picture, can improve the precision of prediction of the motion compensation of B picture, and obtain the raising of code efficiency.Particularly, these methods transmit or the Data processing of storing moving image data of great use.

Claims (4)

1. one kind is used for determining the being included in forward motion vector of the current block in the current picture and the motion vector of backward motion vector is determined method at the direct mode coding, and described motion vector determines that method comprises:

Specify an altogether position piece, described altogether position piece be when being included in the coding current block as backward with reference in the picture of picture and be in piece with described current block same position;

Carry out motion vector and be 0 bi-directional predictive coding, thereby the forward motion vector that will be used for the coding current block and backward motion vector are to indicate as forward direction with reference to the motion vector of the piece of the picture position that comprise, identical with current block of picture with indicate to be used as the motion vector of the piece of position that comprise, identical with current block in the backward picture with reference to picture.

2. one kind is used for determining the being included in forward motion vector of the current block in the current picture and the motion vector of backward motion vector is determined method in the direct mode decoding, and described motion vector determines that method comprises:

Specify an altogether position piece, described altogether position piece be when being included in the decoding current block as backward with reference in the picture of picture and be in piece with described current block same position;

Carry out motion vector and be 0 bi-directional predicted decoding, thereby the forward motion vector that will be used for the decoding current block and backward motion vector are to indicate as forward direction with reference to the motion vector of the piece of the picture position that comprise, identical with current block of picture with indicate to be used as the motion vector of the piece of position that comprise, identical with current block in the backward picture with reference to picture.

3. one kind is used for determining the being included in forward motion vector of the current block in the current picture and the motion vector of backward motion vector is determined equipment at the direct mode coding, and described motion vector determines that equipment comprises:

Can operate the unit that is used to specify common position piece, described altogether position piece be when being included in the coding current block as backward with reference in the picture of picture and be in piece with described current block same position;

Can operate for carrying out motion vector is 0 bi-directional predictive coding, thereby the forward motion vector that will be used for the coding current block and backward motion vector are to indicate as forward direction with reference to the motion vector of the piece of the picture position that comprise, identical with current block of picture with indicate to be used as the unit of the motion vector of the piece of position that comprise, identical with current block in the backward picture with reference to picture.

4. one kind is used for determining the being included in forward motion vector of the current block in the current picture and the motion vector of backward motion vector is determined equipment in the direct mode decoding, and described motion vector determines that equipment comprises:

Can operate the unit that is used to specify common position piece, described altogether position piece be when being included in the decoding current block as backward with reference in the picture of picture and be in piece with described current block same position;

Can operate for carrying out motion vector is 0 bi-directional predicted decoding, thereby the forward motion vector that will be used for the decoding current block and backward motion vector are to indicate as forward direction with reference to the motion vector of the piece of the picture position that comprise, identical with current block of picture with indicate to be used as the unit of the motion vector of the piece of position that comprise, identical with current block in the backward picture with reference to picture.

Images