TWI258991B - Motion vector derivation method - Google Patents

Abstract

When an object macroblock pair (hereinafter MBP) to be coded is coded in a frame structure referring to frame preceding in order of display time, in a mode selecting unit (1109), as for a neighbouring MBP coded in a field structure, when both of 2 macroblocks (hereinafter MB) refer to a frame with the smallest first reference index, motion vector of an macroblock to be coded is calculated by relating a the MBP to an average value of a motion vector of a MB as a top field of a MBP and a motion vector of a MB as a bottom field. When there is a MB referring to a frame containing the first reference index which is not the smallest, a motion vector of an object MB to be coded is calculated by substituting ""0"" into the MB.

Description

1258991 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明The encoding method and the decoding method are not limited to a plurality of images in which the display time is already encoded, a plurality of images in the rear in the chronological order, or a plurality of images in the front and the rear in chronological order. For example, the method of predictive coding is performed. BACKGROUND OF THE INVENTION The encoding of a general moving image is compressed by reducing the length of the time direction and the spatial direction. Therefore, in the inter-picture predictive coding aiming at reducing the verbosity of time, the detection and motion compensation in the block unit are referred to the front or rear image, and the obtained predicted image and the existing image are obtained. The difference 値 of the image is encoded. H. 26L of the current standard motion picture coding method proposes an image in which only intra-picture predictive coding (hereinafter referred to as an I picture) and an inter-picture predictive coding with reference to a 1-frame picture are proposed. (hereinafter referred to as p image), and further refer to two frames of images in front of the display time sequence or two frames of images in the rear in the display time order or two frames of images in the front and rear in order of display time. An image for predictive coding between images (hereinafter referred to as a B image) is performed. Fig. 1 is a view showing an example of a reference relationship between each image in the above-described moving image encoding method and an image referred to here. The image 11 is subjected to intra-image predictive coding without a reference image, and the image 10 is referred to as P7 in front of the display time sequence to perform pre-image pre-image processing, and to describe the code. Further, the image B6 refers to two images in front of the display time sequence, and the image B12 refers to two images in the rear in the display time sequence. The image B18 refers to each of the front and rear in the display time sequence. The image is inter-image predictively encoded. 5 There is a direct mode in the prediction mode of the two-direction prediction in which the inter-image predictive coding is performed in each of the front and rear images in the chronological order. The direct mode does not cause the coding target block to directly have a motion vector, but calculates the motion compensation for actual motion by referring to the motion vector of the block at the same position in the recently coded image in the chronological order. 〇 = one motion vector to make a predicted image. Fig. 2 is a view showing an example of the case where the encoded image referenced by the motion vector is determined in the direct mode, and the motion vector of only one frame image in front of the display time order is referred to. In the same figure, the "P" indicated by the line segment in the vertical direction is independent of the image type and represents only the image. In the same figure, the 15' image P83 is used as an image of the conventional encoding target, and the image P82 and the image P84 are used as reference images for two-direction prediction. When the image P83 is to be coded as the block MB81, the motion vector of the block MB81 is the motion vector of the block MB 82 at the same position of the image P84 after the encoding is completed. Decide. Since the motion vector of the block 20 MB82 has only one of the motion vectors MV81, the two required motion vectors MV82 and motion vectors MV83 are directly related to the motion vector MV81 according to Equations 1(a) and (b). The time interval TR81 is calculated by applying a calibration. • · · Equation 1 MV82 = MV81/TR81 X TR82 7 1258991 Figure 3 (a) shows the spatial prediction method using the conventional direct mode, and when the B image refers to the image in front of the display chronological order An example of a vector prediction method. In the same figure, P denotes a P image, B denotes a B image, and a number added to the image type of the right 4 image indicates a sequential code in which each image is 5 encoded. Here, it is explained that the macroblock to which the oblique line is added to the image B4 is set as the encoding target. In the case where the motion of the coding target macroblock is calculated using the spatial prediction method of the direct mode, first, three encoded macroblocks (dashed line portions) are selected from the periphery of the coding target macroblock. The description of the selection method of the three surrounding macroblocks is omitted here. After the coding, the motion vectors of the 10 macroblocks have been calculated and are maintained. This motion vector is a situation in which macroblocks in the same image are also obtained by referring to different images in each macroblock. The three macroblocks in the periphery are referred to by reference to the image, respectively, by the reference index of the reference image used to encode each macroblock. The details of the reference index will be described in the next section. For example, for the coding target macroblock shown in FIG. 3(a), three surrounding macroblocks are selected, and the motion vectors of each encoded macroblock are respectively set as a motion vector a, a motion vector b, and Move vector c. Here, the moving vector a and the motion vector b are obtained by referring to the P image whose image number 11 is "11", and the motion vector C is obtained by referring to the P image whose image number 11 is "8". In this case, among the motion vectors a, b, and c, the motion vectors a and b that refer to the motion vector of the image in which the target image is displayed at the closest distance by the display time interval become the coding target macroblock. Motion vector candidates. At this time, the motion vector c is regarded as "〇", and the center 値 among the three of the motion vector a, the motion vector b 1258991 玖, the fluttering - and the motion vector C is selected and set as the coding target macro block. Move the vector. However, in the encoding method such as MPEG-4, the interlaced field structure can be used to encode the macroblocks in the image, or the interleaving frame structure can be used for encoding. Therefore, in the case of MPEG-4 in the frame of reference frame 1, a case where a macroblock block coded by the field structure and a macroblock block coded by the frame structure are generated may be generated. In this case, if the three macroblocks around the coding object macroblock are the same as the coding object macroblock, the spatial prediction method of the direct mode can be used to derive an encoding 10 without problems. The motion vector of the object macro block. That is, the case where the coding target macroblock encoded in the frame structure is also coded in the frame structure, or the coding target macroblock in the field structure is also encoded in the field structure. The former situation is as explained. Moreover, the latter case corresponds to the top field of the coding object macroblock by using three 15 motion vectors corresponding to the top field of the three surrounding macroblocks, and the bottom field corresponding to the coding object macroblock is used by Corresponding to the three motion vectors of the bottom fields of the three macroblocks, the motion vectors of the coding target macroblocks can be derived for the respective top and bottom fields by the foregoing method. However, in the case of the temporal prediction of the direct mode, when the block in which the inter-picture 20 prediction coding is performed is compensated by the direct mode, when the block in which the motion vector is referred to belongs to the B picture as shown in FIG. 1 Since the aforementioned block has a plurality of motion vectors, there arises a problem that the motion vector calculated by the scaling of Equation 1 cannot be directly used. Also, since the division calculation is performed after the calculation of the motion vector, the accuracy of the motion vector ( (e.g., 1/2 10 1258991, WW-pixel or 1/4 pixel accuracy) is inconsistent with the predetermined accuracy. In the case of spatial prediction, when any of the coding object macroblock and the surrounding macroblock is encoded in a different structure, there is no provision for any of the structures of the coding object macroblock in the field structure and the frame structure. The encoding, 5' also does not specify a method of selecting a motion vector of a coding target macroblock from among the motion vectors of the field structure coder and the surrounding macroblock of the frame structure coder. SUMMARY OF THE INVENTION [Invention 3] It is an object of the present invention to provide a motion vector prediction method which is provided in a direct mode in a case where a block which is encoded by a field structure or a block which is coded by a frame structure is mixed. It also has a motion vector prediction method with good accuracy in the spatial direction. In order to achieve the above object, the motion vector calculation method of the present invention derives a motion vector calculation method for a motion vector of a block to constitute an image of a moving image from a calculated motion vector of a peripheral block of 15 In the case where the block of the object is encoded or decoded in a frame structure, for the peripheral block coded or decoded by the field structure, the motion vector of the block of the top field and the motion vector of the block of the bottom field are used, The movement vector of the block to be derived is derived by the motion vector 20 of a certain condition, and in the case where the block to be derived is encoded or decoded by the field structure, 'for the peripheral block coded or decoded by the frame structure, use In the motion vector of the two blocks adjacent to each other, the movement satisfying the predetermined condition is derived as the motion vector of the block to be derived, and thus the block to be exported and the surrounding block of 1258991 In the case where the coding or decoding structure is different, it can also be used in the motion vectors of the two blocks of each peripheral block to satisfy certain conditions. The motion vector is derived as a motion vector of the block to be derived. Further, in the above-described moving vector calculation method of the present invention, the predetermined number of five pieces is the case where the block of the derivation target is encoded or decoded in a frame structure, and the movement vector of the block of the top field and the aforementioned bottom field are The image in which the motion vector of the block is referred to separately is a condition in which an image of the same reference index is included in the image that can be referred to, and in the case where the block to be derived is encoded or decoded by the field structure, the foregoing The image in which the moving blocks of the two adjacent blocks are respectively referred to are images having the so-called image of the same reference index in the image that can be referred to. Furthermore, in the above-described motion vector calculation method of the present invention, when both of the motion vectors of the two blocks satisfy the predetermined condition, the average of the motion vectors of the two blocks may be used as the guide. 15 The movement vector of the block of the object. In addition, this manual incorporates the previous Japanese invention patent application "Special Wishes 2002-118598", "Special Wishes 2002-121053", "Special Wishes 2002-156266", "Special Wishes 2002-177889", "Special Wishes 2002- 193027", "Special Wishes 2002-204713", "Special Wishes 2002-262151", "Special Wishes 2002-20 290542", "Special Wishes 2002-323096", US Applications "60/378643" and "60/378954" Content. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic diagram for explaining the reference relationship of the image of the conventional example. 12 125899 ————— —

P Fig. 2 is a schematic view for explaining the action of the direct mode of the conventional example

V 〇 Fig. 3(a) shows an example of a motion vector -5 prediction method in the case of a B-image reference time using the spatial prediction method of the conventional direct mode. Fig. 3(b) shows an example of a reference image list in which each encoding target image is created. Fig. 4 (A) and (B) are explanatory diagrams of an image number and a reference index. Fig. 5 is a view showing a format of an image coding signal No. 10 format constituted by a conventional image coding apparatus. Fig. 6 is a block diagram for explaining the coding operation of the embodiment 1 and the embodiment 2 of the present invention. Fig. 7 is a schematic diagram for explaining the action of the block in which the reference vector is referred to in the direct mode with the reference to display the two motion vectors ahead in the chronological order. Fig. 8 (a) and (b) are images for comparing the display order and the coding order. Fig. 9 is a view for explaining a block in which a motion vector is referred to in the direct mode. # A mode diagram having an action of referring to the two motion vectors in the rear in the chronological order. Fig. 10 (a) and (b) are pattern diagrams for comparing the image reference relationship between the display order and the encoding order. Figure 11 is a block diagram for explaining the decoding operation of the embodiment 5 and the embodiment 6 of the present invention. 13 1258991 —, -ffiOTT - Fig. 12 is a schematic diagram for explaining the operation of the reference moving vector in the direct mode with reference to the two moving vectors in the chronological order. Fig. 13 is a schematic diagram for explaining the operation of the block 5 which refers to the motion vector in the direct mode with reference to the case of displaying the two motion vectors in the rear in the chronological order. Fig. 14 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. 10 Fig. 15 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the two mobile vectors in the chronological order. Fig. 16 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. Fig. 17 is a schematic diagram for explaining the operation of the block in which the motion vector is referred to in the direct mode with reference to the two motion vectors in the chronological order. Fig. 18 is a schematic diagram for explaining the operation of the block 20 which refers to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. Fig. 19 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. 14 1258991 Rose τ invention description + Fig. 41 (b) shows the order in which the arrangement of the frames shown in Fig. 41 (a) is replaced by the coding order. Fig. 42 is a view showing the construction of a reference image list for explaining the first embodiment. 5 Fig. 43(a) shows an example of a motion vector calculation sequence using a prediction method in a direct mode space in which a macroblock block paired with a field structure code and a macroblock block coded by a frame structure are mixed. flow chart. Fig. 43(b) shows an example of the arrangement of the peripheral macroblock pairings to which the present invention is applied in the case where the coding target macroblock pairing is coded in a frame structure. Fig. 43(c) shows an example of the arrangement of the peripheral macroblock pairing to which the present invention is applied in the case where the encoding target macroblock pairing is coded by the field structure. Fig. 44 is a diagram showing the data composition of the macroblock block pairing in the frame structure coding case and the data structure of the macroblock block in the case of the field structure coding. Fig. 45 is a flow chart showing a more detailed sequence of steps of step S302 shown in Fig. 43. Fig. 46 is a view showing the relationship between the reference field index and the reference frame index. 20 Fig. 47 is a flow chart showing a more detailed sequence of steps of step S303 shown in Fig. 43. Fig. 48 (a) and (b) are diagrams showing the positional relationship between the encoding target block pairing and the peripheral macroblock pairing in the first embodiment. Fig. 49 (a) and (b) show coding for explaining the i-th embodiment. 17 1258991 发明, description of the invention Fig. 3(b) shows an example of a reference image list for creating each coding target image. The reference image list 1 shown in FIG. 3(b) is displayed in front of and behind the time centering on one b image, and indicates a peripheral image of the stomach of the B image, and a map of the images. The image type, the image number u, the i-th folder 5, and the second reference index 13. The image number U indicates, for example, the number of each image in the encoding order. The first reference index 12 indicates a first index for the relative positional relationship of the peripheral image of the stomach image, and is used, for example, in the encoding target image to display the chronological order in the front image. The list of the first reference index 12 is referred to as "reference index list 10 〇 (HstO)" or "first reference index list". Also, the reference index is also referred to as a relative index. In the reference image list 10 of FIG. 3(b), after the first reference index 12, first, for the reference image having the display time earlier than the encoding target image, the encoding target image is allocated in display time order from The more recent order is increased by "〇" by an integer of "1". When all the reference image assignments having the display time earlier than the image of the encoding target image 15 are increased by "1" from "〇", the reference image having the display time later than the encoding target image is next. The encoding object image distribution is continuous by the order in which the display time sequences are relatively close. The second reference index 13 indicates a second index of the relative positional relationship with respect to the peripheral image of the image to be encoded, and is mainly used, for example, for the index of the encoding target image to be displayed in the chronological order in the rear image. The list of the second reference index is called "reference index list l (listl)" or "second reference index list". Further, after the second reference index I2, first, for the reference image having the display time later than the encoding target image, the image is assigned to the encoding target 19 1258991 in order of display time from the closer order by "〇". Raise the integer 「 of "1". When all the reference image assignments having the display time later than the encoding target are increased by "1" from "0", the reference image having the display time earlier than the encoding target image is used in the encoding pair. 5 Image image assignment is continuous by the order in which the display time sequence is closer. Therefore, when the reference image list 10 is viewed, the first reference index 12 and the second reference index 13 can be known, and the reference image whose reference index is smaller is closer to the encoding target image in the display time order. The above has explained the allocation method of the number in the initial state of the reference cable. However, the method of referring to the number of the index can be changed in the image unit or the slice unit. Referring to the method of assigning the numbers of the reference numerals, for example, it is also possible to assign a small number to the image in the order of display time, but such a reference index, for example, refers to an image in which the images are separated in time order. It is used in situations where coding efficiency is improved. That is, the reference index in the block is represented by the variable length codeword, and the code of the shorter code length is allocated from 15 when the recovery is smaller, so that the image with higher coding efficiency is improved relative to the reference. A small reference index reduces the amount of code of the reference index and further improves the coding efficiency. Fig. 4 is an explanatory diagram of an image number and a reference index. The fourth chart axe: The example of the reference image list indicates the B image at the center (the dotted line indicates the test). 2) The reference image used for encoding and its image number and reference index. Fig. 4(A) shows a case where the reference index is allocated by the allocation method of the reference index in the initial state explained in Fig. 3. Fig. 5 is a view showing a format of an image coded signal format constituted by a conventional image coding apparatus. Picture indicates the encoded signal of 1 image size, 20 1258991

Header indicates the header coded signal before the head of the picture, Block1 indicates the coded signal of the block formed by the direct mode, and Block2 indicates the coded signal of the block formed by the interpolation prediction other than the direct mode, RidxO and Ridxl respectively indicate The first reference index and the second reference index, MV0 and 5 MV1 respectively indicate the first motion vector and the second motion vector. The coding block Block 2 is used to indicate the two reference pictures used for the interpolation, and the reference indices RidxO and Ridxl are set in the coded signal in this order. Further, the first motion vector MV0 and the second motion vector MV1 of the coding block Block 2 are encoded in the coded signal of the coding block Block2 in this order. In the end, you can use the PredType to determine which of RidxO and Ridxl is used. Further, the image to be referred to by the first motion vector MV0 (first reference image) is indicated by the first reference index RidxO, and the image to be referred to by the second motion vector MV1 is indicated by the second reference index Ridx1 (second reference) image). For example, RidxO and Ridx1 can be used in the case of referring to the image in the two directions of the motion vectors MV0 and MV1, and it is indicated that the vector can be used in the case where the image is referred to in one of the motion vectors MV0 or MV1. Refer to RidxO or Ridxl of the index, and in the case of direct mode, RidxO and Ridxl are not used. The first reference image is specified by the first reference index, and is generally a picture of FIG. 20 having a display time earlier than the image to be encoded. The second reference picture is specified by the second reference index, and generally has a comparison target. The image is displayed later in the image. However, it can be seen from the example of the method of the reference index of FIG. 4 that the first reference image system has an image at a display time later than the image to be encoded, and the second reference image has a comparison target. The case where the image is displayed in front of the image. The first reference 21 1258991 — — —

The index RidxO indicates the reference index of the first reference image of the first motion vector MVO of the reference block Block2, and the second reference index Ridx1 indicates the reference of the second reference image of the second motion vector MV1 of the reference block Block2. , according to the index. 5 On the other hand, by using the buffer control signal in the encoded signal (Figure 5)

The RPSL in the header and the explicit indication can arbitrarily change the allocation of the reference image to the index. By changing the allocation, the reference image of "0" can be set to an arbitrary reference image in the second reference index. For example, the assignment of the reference index to the image number can be changed as shown in Fig. 4 (B). . In this case, since the assignment of the reference image to the reference index can be arbitrarily changed, and the assignment of the reference image to the reference index is generally changed, the reference image is increased in comparison with the coding efficiency by selection. As if a smaller reference index is to be allocated, the coding efficiency can be improved by setting the motion vector whose reference index of the reference image referenced by the motion vector to the minimum is set to the motion vector used in the direct mode. (Implementation Mode 1) A moving picture coding method according to Embodiment 1 of the present invention will be described with reference to a block diagram shown in Fig. 6. The moving image to be encoded is input to the frame memory 101 in units of images in chronological order, and is arranged in the order in which the encoding is performed. Each image is called a block. For example, it is divided into groups of 16 X vertical and 16 pixels horizontally, and the subsequent processing can be performed in block units. The block read from the frame memory 101 is input to the motion vector detecting unit 106. Here, it is explained that the encoded image stored in the frame memory 105 is used as an image to detect the motion vector of the block to be encoded as the image to be decoded. At this time, the mode selection unit 107 refers to the motion vector obtained by the motion vector detecting unit 106 or the motion vector used in the encoded image stored in the motion vector storage unit 108 to determine the most appropriate prediction mode -5. The prediction image obtained by the mode selection unit 1〇7 and the prediction image determined by using the motion vector used in the prediction mode are input to the difference calculation unit '109, and the prediction residual is generated by taking the difference from the block to be coded. The difference image is encoded by the prediction residual encoding unit 102. Further, the motion vector used in the prediction mode obtained by the modular selection unit 107 is used for encoding the image or block of the subsequent 10, and is stored in the motion vector storage unit 108. The above processing flow is an operation for selecting inter-picture predictive coding, and switching with intra-picture predictive coding can be performed only by the switch 111. Finally, the code column generating unit 103 performs variable length coding on the control information output from the motion vector and the image information output from the prediction residual encoding unit 102 to generate a final output 15 code sequence. The above shows the flow of the encoding. However, the processing of the motion vector detecting unit 106 and the mode selecting unit 107 will be described in detail below. The detection of the motion vector is performed in the field of each block or each partition. For the image to be encoded, the encoded image located in front of and behind the & 2 is used as the reference image in the display time sequence, and the prediction field in the image is determined by the decision to be the most appropriate. The motion vector and the prediction mode of the position are used as the predicted image. ~ Refer to the two frames of images in front and back in order to display one of the two directions for inter-picture predictive coding. There is a direct mode 23 1258991 wvmmm —————— . The direct mode does not directly make the block of the coding object have a motion vector, but calculates the two motion vectors for actually compensating by referring to the motion vector of the same position block in the coded image in the chronological order. And make a predicted image. 5 Fig. 7 illustrates the actor in the case where the block of the motion vector referred to in the direct mode for determining the motion vector has a reference to display two motion vectors of the two frames in front of the chronological order. The image P23 is an image to be encoded now, and the image P22 and the image P24 are used as reference images to perform two-direction prediction. When the block 10 to be coded is set as the block MB21, the two motion vectors that are deemed to be necessary at this time use the second reference image specified by the second reference index after the encoding is completed. The image P24 is determined by the motion vector of the block MB22 at the same position. The block MBU has two motion vectors MV21 and a motion vector MV22 as motion vectors. Therefore, it is impossible to directly calculate the required two motion vectors MVU and motion vectors by applying the same direct scaling as Equation 1. MV24. Therefore, as in Equation 2, on the motion vector to which the application is scaled, the motion vector MV_REF is calculated from the average of the two motion vectors of the block MB22, and the time interval TR_REF at this time is calculated from the average 同样. The motion vector MV23 and the motion vector MV24 are calculated for the motion vector MV-20 REF and the time interval TR_REF by applying scaling according to Equation 3. At this time, the time interval TR21 indicates the time interval from the image P24 to the image PM, that is, the image to which the motion vector MV21 refers, and the time interval TR22 indicates the time interval until the image to which the motion vector MV22 refers. . Also, the time interval TR23 24 1258991 ------------------------------------------- - The time interval until the image referred to by the motion vector MV23 is indicated, and the time interval TR24 indicates the time interval until the image referred to by the motion vector MV24. The time interval between the images can be determined based on, for example, information indicating the display time or display order of each image, or based on the information difference. Further, the image to be encoded in the example of Fig. 7 refers to an adjacent image, but the same processing can be performed even when referring to a non-adjacent image.

MV_REF= (MV21 + MV22)/2 · · · · Equation 2 (4) 10 TR_REF- (TR21 + TR22)/2 · · · · Equation 2 (b) MV23 = MV—REF/TR—REF x TR23 (a) MV24 - -MV_REF/TR_REF x TR24 15 (b)

As described above, the above embodiment is shown in the direct mode, in the case where the block of the reference motion vector has a reference to display a majority of the motion vector of the image in front of the time sequence, a majority of the motion vector is used to generate a gravity direction. 4 'Determining the two motion vectors actually used for the motion-to-compensation 20 by applying the calibration, even in the case of referring to the region b image of the motion vector in the direct mode, the direct mode can be used without contradiction. The coding method of inter-picture predictive coding. In addition, when the two motion vectors MV23 in FIG. 7 and the movement to the stomach MV24 are requested, the motion vector 25 1258991 翻, —-MV-REF, and time interval TR_REF for calculating the target to be benchmarked are calculated. On the method of obtaining the average 値 of the motion vector MV21 and the motion vector MV22 and the average 时间 of the time interval TR21 and the time interval TR22, Equation 4 can be used instead of Equation 2. First, as shown in Fig. 4 (4), the motion vector MV21 time interval is scaled in the same manner as the 5 motion vector MV22, and the motion vector MV21 is calculated. The motion vector MV_REF is determined by taking the average of the motion vector MV21' and the motion vector MV22. At this time, the time interval TR_REF is the original local use time interval TR22. Further, the case where the motion vector MV21 is set as the motion vector MV21' and the motion vector MV22 is scaled to the motion vector MV22' can be similarly performed. MV21, = MV21/TR21 X TR22 · · · · Equation 4 (4) MV_REF=(MV21' + MV22)/2 · · · · Equation 4 (b) 15 TR_REF=TR22 · · •• Equation 4 (c) Again, When calculating the two motion vectors MV23 and the motion vector MV24 in FIG. 7, the motion vector MV_REF and the time interval TR_REF, which are objects to be scaled, can replace the 20 average 使用 of the two motion vectors of Equation 2, but The reference of Equation 5 is directly used for the motion vector MV22 and the time interval TR22 of the image P22 having a short time interval with reference to the image P24 of the motion vector. Similarly, the motion vector MV21 and the time interval TR21 of the image P21 having a long reference time interval can be directly used as the motion vector MV_REF time interval TR_REF. By this method, each block belonging to the image P24 to which the mobile 26 1258991 vector is to be referred to can perform motion compensation by memorizing only one of the two motion vectors, and thus the capacity of the motion vector memory portion of the encoding device can be realized. The suppression is small. MV_ REF 2: MV22 • · •• 式 5 5 (a) TR _REF 2 TR22 • · · · Equation 5 (b) MV. A REF 2: MV21 • · · · Equation 6 (a) 10 TR _REF = TR21 • · · • Equation 6 (b) In addition, when calculating the two motion vectors MV23 and MV24 of Fig. 7, as the motion vector MV_REF and time interval TR to be subjected to scaling A REF can be used instead of the 15 値 of the two motion vectors of Equation 2, and the motion vector of the previous image with reference to the sequence to be encoded is directly used. Fig. 8(a) is a reference relationship showing an arrangement pattern of images to be displayed as moving images in Fig. 7, and Fig. 8(b) shows the order in which the frame memory 101 in Fig. 6 is to be encoded. An example of a replacement arrangement. Further, the image P23 indicates an image encoded by the direct mode, and the image 24 indicates an image with reference to the motion vector at 20 o'clock. When the arrangement is replaced as in the case of FIG. 8(b), since the motion vector of the previous image to be encoded is directly used, as shown in Equation 5, the motion vector MV22 and the time interval TR22 can be directly applied as The motion vector MV_REF and the time interval TR_REF. Similarly, the motion vector 27 1258991 of the image following the order to be encoded can also be used directly. In this case, as shown in Equation 6, the motion vector MV21 and the time interval TR21 can be directly applied as the motion vector MV_REF and the time interval TR_REF. By this method, each block belonging to the image p24 to be referred to the motion vector can realize the shifting motion compensation by memorizing only the state of one of the two motion vectors. Therefore, the capacity of the motion vector memory portion of the encoding device can be realized. The suppression is small. In the present embodiment, the calibration is performed for the distance of the moving vector to be referenced, and the motion vector used in the direct mode is calculated. This can also be referred to the moving vector to be referred to. Usually 1 计算 multiple times calculation. Here, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. In addition, when calculating the motion vector MV_REF in Equation 2(a) or Equation 4(b), after calculating the right side of Equation 2(a) or Equation 4(b), it is also possible to obtain a predetermined motion vector accuracy (for example, 1). The movement vector of /2 pixel accuracy is 〇.5 like 15 units. As for the accuracy of the motion vector, it is not limited to 1/2 pixel accuracy. And the accuracy of this motion vector is determined, for example, by a block unit, an image unit, and a sequence unit. Further, when the motion vector MV23, the motion vector MV24, and the motion vector MV21 are calculated in Equation 3(a), Equation 3(b), or 4(a), the hn equations 3(a) and 3(b) may be used. Or the right side of 4(a) to make a predetermined 20 motion vector accuracy. (Implementation 2) The outline of the encoding process according to Fig. 6 is exactly the same as that of the embodiment 1. Here, the action predicted for the two directions of the direct mode will be described in detail using FIG. 28 1258991 SH, SHMT — — — Figure 9 illustrates the action in the direct mode to determine the motion vector and the referenced block has a reference to display the two motion vectors in the chronological order. The image P43 is an image to be encoded now, and the image P42 and the image P44 are used as reference images to perform two-direction prediction. When the block to be coded is set to block MB41, the two motion vectors that are deemed necessary at this time are used in the encoded back reference image (the second reference image specified by the second reference index). The image P44 is determined by the motion vector of the block MB42 at the same position. Since the block MB42 has two motion vectors such as the motion vector MV45 and the motion vector MV46, it is not possible to directly apply scaling to calculate the two required motion vectors MV43 and motion vectors MV44 in the same manner as in Equation 1. As shown in Fig. 7, the motion vector MV_REF is determined from the average 値 of the two motion vectors of the block MB42 on the motion vector as the application calibration, and the time interval TR at this time is determined by the average 値. REF. The motion vector MV43 and the motion vector MV44 are calculated by applying scaling to the motion vector MV 15 —REF and the time interval TR_REF according to Equation 8. At this time, the time interval TR45 indicates the interval from the image P44 to the image P45, that is, the time until the image referred to by the motion vector MV45, and the time interval TR46 indicates the time until the image referenced by the motion vector MV46. interval. Further, the time interval 20 TR43 represents the interval from the time until the image referred to by the motion vector MV43. The time interval TR44 indicates the interval to the time until the image referenced by the motion vector MV44. The time interval between the images is the same as that described in the embodiment 1, and is determined, for example, based on information indicating the display time or display order of each image, or information difference thereof. Further, in the example of Fig. 9, the image to be encoded is referred to as an adjacent image, and can be similarly processed when referring to a non-adjacent image. MV_REF= (MV45 + MV46)/2 (a) • · · · Equation 7 5 TR—REF = (TR45 + TR46)/ 2 (b) • · •• Equation 7 MV43= -MV_REF/TR_REF x TR43 (a) • 式8 MV44=MV_REF/TR_REF x TR44 • • • • Equation 8 10 (b) As described above, an encoding method is disclosed in which the block of the direct mode reference motion vector has a reference. In the case where the chronological order is in the majority of the motion vectors of the rear image, a motion vector is generated using the aforementioned majority motion vectors, and the two motion vectors used for the motion compensation are actually determined by applying the scaling. The direct mode can achieve the encoding method using the inter-picture predictive coding of the direct mode without contradiction even in the case where the block of the reference motion vector belongs to the B picture. When the motion vector MV43 and the motion vector MV44 in the ninth figure are requested, in order to calculate the motion vector MV 20 — REF and the time interval TR — REF of the target to be scaled, the motion vector MV 45 and the motion are acquired. For the method of averaging the vector MV46 and the average 値 of the time interval TR45 and the time interval TR46, Equation 9 can be used instead of Equation 7. First, the motion vector MV46 is calculated by scaling the motion vector MV46 as shown in Fig. 9(a) so that the time interval is the same as the motion vector MV45. The motion vector MV_REF is determined by averaging the motion vector MV46' and the motion vector MV45. At this time, the time interval TR_REF is the original local use time interval TR41. Further, the case where the motion vector MV46 is scaled as the motion vector MV46' and the motion vector MV45 is scaled as the 5 motion vector MV45' can be handled in the same manner. MV46,-MV46/TR46 x TR45 · · · · Equation 9 (a) MV_REF=(MV46' + MV45)/2 · · · • Equation 9 (b) 10 TR_REF-TR45 · · · · Equation 9 (c) When calculating the two motion vectors MV43 and the motion vector MV44 in FIG. 9, the motion vector MV_REF and the time interval TR_REF, which are the target of the scaling, can be used instead of Equation 7. The average 値 of the amount of 15 is used, and the motion vector MV45 and the time interval TR45 of the image P45 having the short reference time interval are directly used for the image P44 of the motion vector to be referred to. Similarly, the motion vector MV46 and the time interval TR46 of the image P46 separated by the time interval to be referred to in Equation 11 can be directly used as the motion vector MV_REF and the time interval TR_REF. By this method, each block belonging to the image P44 to be referred to the motion vector can realize motion compensation by memorizing only one of the two motion vectors, so that the capacity of the motion vector memory portion of the encoding device can be suppressed. Get small. MV_REF = MV45 · · · · Equation 10 1258991 玖, — TR REF=TR45 • · · · Equation 10 (b) MV_ _REF=MV46 • · •• Equation 11 (4) TR One REF=TR46 • · · · 11 (b) In addition, when calculating the two motion vectors MV43 and the motion vector MV44 in Fig. 9, the motion vector MV_REF and the time interval TR_REF which are the target of the scaling are used instead of the equation 7. The two move to an average of 10 quantities, and directly use the motion vector of the image with the first order to be encoded. Fig. 10(a) shows the reference relationship of the arrangement of the images in the order of the moving image display as shown in Fig. 9, and Fig. 10(b) shows the frame memory 101 in Fig. 6 in the order of encoding. Replace one of the examples. Further, the image P43 indicates an image encoded by the direct mode, and the image 15 P44 indicates an image with reference to the motion vector at this time. When the arrangement is replaced by FIG. 10(b), the motion vector MV46 and the time interval TR46 can be directly used as the motion vector MV_REF by directly using the motion vector of the image whose order is to be encoded first. Time interval TR_REF. It is also possible to directly use the motion vector of the image whose reference encoding order is followed. In this case, as shown in Equation 10, the motion vector MV45 and the time interval TR# can be directly applied as the motion vector MV_REF and the time interval TR_REF. According to this method, each block belonging to the image P44 to be referred to the motion vector can realize motion compensation by memorizing only one of the two motion vectors, so that the capacity of the motion vector memory unit of the encoding device can be suppressed. small. 32 1258991 莰,—————— Also, when the direct mode is used to determine the motion vector and the reference image is referenced to display the two motion vectors of the second frame image in the chronological order, the request may also be obtained. The two motion vectors MV43 and the motion vector MV44 are set to "〇" to perform motion compensation. By this method, each block belonging to the 5 image P44 to which the motion vector is to be referred to can eliminate the _3 memory vector, so that the capacity of the motion vector memory unit of the encoding device can be suppressed small, and the calculation can be omitted. The processing of moving vectors. Further, in the case where the image referred to in the direct mode for determining the motion vector has the reference to display the two motion vectors of the two frames of the rear two images, the reference motion vector may be prohibited from being applied only to the direct mode. Predictive coding. In the case where the image P44 of FIG. 9 refers to the two frames of images in the rear in order of display time, it is known that the possibility of being associated with the image in front in the display time order is low, so that the direct mode is prohibited. Other prediction methods can produce more accurate predicted images. Further, in the present embodiment, the case where the distance between the images is used for the motion vector to be referred to is scaled, thereby calculating the motion vector used in the direct mode, which may also be referred to. The vector is given a constant multiple of ten. Here, the constant used in the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. 20 'In the case of Equation 7(a) or Equation 9(b), when the motion vector MV_REF is calculated, the right side of Equation 7(a) or Equation 9(b) can also be used to accurately determine the motion vector. Within. As for the accuracy of the motion vector, there are 1/2, 1/3 pixel, and 1/4 pixel precision. And the accuracy of this motion vector is determined, for example, by a block unit, an image unit, and a sequence unit. Further, when the motion vector MV43 'moving vector MV44 and the motion vector MV46' are calculated in Equation 8(a), Equation 8(b), and 33 1258991 9(a), Equation 8(a) and Equation 8(b) can also be calculated. ), or the right side of 9(a) to make a predetermined motion vector accuracy. (Implementation 3) 5 A moving picture coding method according to Embodiment 3 of the present invention will be described with reference to the block diagram of FIG. However, the code train generated by the dynamic image coding method of the mode 1 is input. First, various information such as prediction mode, motion vector information, and prediction residual coded data are extracted from the input code sequence by the code column parser 601. The prediction mode/moving vector decoding unit 608 outputs the prediction mode or the motion vector information, and outputs the prediction residual coding unit 602 to the prediction residual decoding unit 602. The prediction mode/moving vector decoding unit 608 performs decoding of the prediction mode and decoding of the motion vector used in the prediction mode. When the motion vector is decoded, the decoded motion vector 15 stored in the motion vector memory unit 605 is used. The decoded prediction mode and the motion vector are output to the motion compensation decoding unit 6〇4. Further, the motion vector storage unit 605 is stored in order to cause the decoded motion vector to be used for decoding the motion vector of the subsequent block. The motion-to-compensation decoding unit 604 uses the decoded image of the decoded image stored in the frame memory 603 as a reference image, and generates a predicted image based on the input prediction mode or motion vector information. The predicted image input addition unit calculation unit 606 thus generated generates a decoded image by performing addition of the prediction residual image generated by the prediction residual decoding unit 602. The above embodiment is for the operation of the code sequence for inter-picture predictive coding, but the switching of the decoding process for the code sequence for intra-picture predictive coding can be performed by the switch 607. 34 1258991 玖, - The above shows a summary of the flow of decoding, and the details of the processing of the move-to-compensation decoding unit 604 will be described below. The motion vector information is appended to each block or to the area of each partition. The decoded image of the image to be decoded in the front or the last 5 positions in the display time order is used as a reference image and is generated from the image by the decoded motion vector. Moving to the compensated predicted image 〇 refers to displaying the image of each frame in the front and rear in the chronological order as one of the two direction predictions for inter-picture predictive coding. Directly inputting a code sequence having no motion vector in the block of the decoding target in the direct mode, and thus calculating the actual operation by referring to the motion vector of the block at the same position in the decoded image in the chronological order. The two motion vectors of the motion compensation are used to create a predicted image. Fig. 7 is a diagram showing the actor in the case where the block for moving the vector in the direct mode and the motion vector of the reference 15 has a reference to display two motion vectors of the two frames in front of the chronological order. The image P23 is an image to be encoded now, and the image P22 and the image P24 are used as reference images to perform two-direction prediction. When the block to be coded is set to block MB21, the two motion vector systems 20 deemed necessary at this time use the second reference image specified by the second reference index after the encoding is completed. The image P24 is determined by the motion vector of the block MB22 at the same position. The block MB22 has two movement vectors MV21 and a motion vector MV22 as the motion vector. Therefore, it is not possible to directly calculate the required movement by applying the same direct scaling as Equation 1 to 35 1258991—one— — — — " ~ MV23 and motion vector MV24. Therefore, as in Equation 2, on the motion vector to be applied to the calibration, the motion vector MV_REF' is calculated from the average of the two motion vectors of the block MB22, and the time interval TR at this time is calculated from the average —. REF. The motion vector MV23 and the motion vector MV24 are calculated for the motion vector MV_'5 REF and the time interval TR_REF by applying scaling according to Equation 3. At this time, the time interval TR21 indicates the time interval from the picture 'P24 to the picture P21, that is, the time to the image referred to by the motion vector MV21'. The time interval TR22 indicates the image to be referred to by the motion vector·MV22. time interval. Further, the time interval TR23 10 represents the time interval until the image referred to by the motion vector MV23, and the time interval TR24 represents the time interval until the image referred to by the motion vector MVM. The time interval between the images can be determined based on, for example, information indicating the display time or display order of each image, or based on the information difference. Further, in the image to be encoded in the example of Fig. 7, the image adjacent to the phase 15 is referred to, but the same processing can be performed even when referring to a non-adjacent image. · As described above, the above embodiment is shown in the direct mode, in the case where the block of the reference motion vector has a reference to display a multi-y-number motion vector of the image in front in chronological order, a majority of the motion vector is used to generate one 3 20 The motion vector 'determines the two motion vectors actually used for the motion direction compensation by applying the calibration, even if the block in the direct mode referring to the motion vector belongs to the B picture, the direct use can be used without contradiction. Pattern-based graphing—the encoding method for inter-picture predictive coding. In addition, when the two motion vectors MV23 in FIG. 7 are moved and the MV2 is shifted to 36 1258991, the motion vector MV_REF and the time interval TR_REF which are the targets of the fixed target are calculated, and the motion vector MV21 is obtained. In the method of moving the vector MV22 and the average 値 and the average 値 of the time interval TR21 and the time interval TR22, Equation 4 5 can be used instead of Equation 2. First, as shown in Fig. 4(a), the motion vector MV21 time interval is scaled in the same manner as the motion vector MV22, and the motion vector MV21' is calculated. The motion vector MV_REF is determined by taking the average of the motion vector MV21' and the motion vector MV22. At this time, the time interval TR_REF is the original local use time interval TR22. Further, the case where the motion vector MV22 is set as the motion vector MV21' instead of the motion vector MV21 and the motion vector MV22 is scaled to the motion vector MV22' can be similarly performed. Further, when calculating the two motion vectors MV23 and the motion vector MV24 of FIG. 7, the motion vector MV_REF and the time interval TR_REF which are the targets to be subjected to the scaling are the 15 averaging 使用 using the two motion vectors instead of Equation 2. On the other hand, the reference vector 5 can be directly used for the motion vector MV22 and the time interval TR22 of the image P22 having a short time interval with reference to the image P24 of the motion vector. Similarly, the motion vector MV21 and the time interval TR21 of the image P21 having a long reference time interval can be directly used as the motion vector MV_REF time interval TR_REF. By this method, each block belonging to the image P24 to be referred to the moving 20 vector can realize motion compensation by memorizing only one of the two motion vectors, and thus the amount of the moving vector memory of the decoding device can be realized. The suppression is small.

Moreover, when calculating the two motion vectors MV23 and the motion vector MV24 of FIG. 7, the motion vector MV REF 37 1258991, which is the object to be scaled, and the time interval TR REF can be substituted for Equation 2 The average 値 of the two motion vectors is used, and the motion vector of the previous image in the order of decoding is directly used. Fig. 8(a) is a reference relationship showing an arrangement pattern of images to be displayed as moving images in Fig. 7, and Fig. 8(b) shows the order of 5 columns of input codes, i.e., to be decoded. An example of the order. Further, the image P23 indicates an image decoded by the direct mode, and the image 24 indicates an image with reference to the motion vector. When considering the order of arrangement in Fig. 8(b), since the motion vector of the previous image in the order of decoding is directly used, as shown in Equation 5, the motion vector MV22 and the time interval TR22 can be directly applied as 10 It is the motion vector MV_REF and the time interval TR_REF. Similarly, the motion vector of the image following the order to be decoded can also be used directly. In this case, as shown in Equation 6, the motion vector MV21 and the time interval TR21 can be directly applied as the motion vector MV_REF and the time interval TR_REF. According to this method, the block 15 of each region belonging to the image P24 to be referred to the motion vector can realize the motion compensation by memorizing only one of the two motion vectors, so that the capacity of the motion vector memory portion of the decoding device can be suppressed. Get small. Moreover, in the present embodiment, the case where the distance between the images is used for the motion vector to be referred to is scaled, and the motion vector for the direct mode 20 is calculated, which may also be referred to. The vector is calculated as a constant multiple. Here, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. (Implementation of the sample dragon 4) The summary of the code processing according to Fig. 11 is the same as the implementation mode 3 38 1258991 玖, the invention description. Here, the action predicted for the two directions of the direct mode will be described in detail using FIG. However, it is assumed that the code sequence generated by the moving picture coding method of the mode 2 is input. Fig. 9 is a view for explaining the operation in the case where the block for reference is used in the direct mode and the block of the reference 5 has two moving vectors which are referenced to display two frames of images behind the time sequence. The image P43 is an image to be decoded now, and the image P42 and the image P44 are used as reference images to perform prediction in two directions. Once the block to be decoded is set to block MB41, the two motion vectors that are deemed necessary at this time are used in the decoded back reference picture (the second reference picture specified by the second reference index). The image P44 is determined by the motion vector of the block MB42 at the same position. This block MB42 has two motion vectors such as a motion vector MV45 and a motion vector MV46. Therefore, it is not possible to directly apply scaling to calculate the required two motion vectors MV43 and motion vectors MV44 in the same manner as in Equation 1. As shown in Fig. 7, the motion vector MV_REF is determined from the average 値 of the two motion vectors of the block MB42 in the motion vector for which 15 is applied for scaling, and the average 値 is used to determine the current 値. Time interval TR - REF. The motion vector MVU and the motion vector MV44 are calculated by applying scaling for the motion vector MV_REF and the time interval TR_REF according to Equation 8. At this time, the time interval TR45 indicates the interval from the 20 image P44 to the image P45, that is, the time until the image referred to by the motion vector MV45, and the time interval TR46 indicates the time until the image to which the motion vector MV46 refers. The interval, time interval TR43, represents the interval to the time until the image referenced by the motion vector MV43. The inter-day interval TR44 is not as long as the image referenced by the motion vector MV44. 39 1258991 The interval between the time of the invention and the description. Further, the image to be decoded in the example of Fig. 9 refers to an adjacent image, and can be handled similarly when referring to a non-adjacent image. As described above, a decoding method is disclosed in which the block of the direct mode reference motion vector has a reference to display a plurality of motion vectors of the image in the chronological order, using the plurality of motion vectors. A motion vector is generated, and two motion vectors for actually using motion compensation are determined by applying scaling, and in the direct mode, even if the block referring to the motion vector belongs to the B image, the collision can be achieved without contradiction. A decoding method using inter-picture prediction decoding in direct mode. When the motion vector MV43 and the motion vector MV44 in the ninth figure are required, the motion vector MV_REF and the time interval TR_REF of the target to be scaled are calculated as the motion vector MV45 and the motion vector MV46. In the method of averaging 时间 and the time interval TR45 and the average 値 of the time interval 15 TR46, Equation 9 can be used instead of Equation 7. First, the motion vector MV46' is calculated by scaling the motion vector MV46 as shown in Fig. 9(a) so that the time interval is the same as the motion vector MV45. The motion vector MV_REF is determined by taking the average of the motion vector MV46' and the motion vector MV45. At this time, the time interval TR_REF is originally used when the local interval is 2 U〇5. Further, instead of scaling the motion vector MV46 as the motion vector MV46' and scaling the motion vector MV45 as the motion vector MV45', the same can be applied.

In addition, when calculating the two motion vectors MV43 and the motion vector MV44 of the brother 9 diagram, the motion vector MV_REF 40 1258991, which is the target of the calibration, -,------------ ----------------- and the time interval TR REF, instead of the average 値 of the two moving vectors using Equation 7, and the moving vector to be referred to as Equation 10 The image P44 directly uses the motion vector MV45 and the time interval TR45 of the image P45 having a short reference time interval. Similarly, the motion vector MV46 and the time interval TR46 of the image P46 separated by the time interval 5 to be referred to as Equation 11 can be directly used as the motion vector MV_REF and the time interval TR_REF. According to this method, each block belonging to the image P44 to be referred to the motion vector can realize motion compensation by memorizing only one of the two motion vectors, so that the capacity of the motion vector memory unit of the decoding device can be suppressed. small. 10, when calculating the two motion vectors MV43 and the motion vector MV44 in FIG. 9, the motion vector MV_REF and the time interval TR_REF as the target of the scaling are used instead of the equation 7. The average of the moving vectors is used, and the moving vector of the image with the first order to be encoded is directly used. Fig. 10(a) shows the reference relationship of the arrangement of the images in the order of the moving image display 15 as shown in Fig. 9, and Fig. 10(b) shows the order of the input code columns, i.e., to be decoded. An example of the order. Further, the image P43 indicates an image encoded by the direct mode, and the image P44 indicates an image with reference to the motion vector at this time. Considering the arrangement order of the first diagram (b), the motion vector MV46 and the time interval TR46 can be directly used as the motion by directly using the motion 20 vector of the image in which the order of decoding is to be directly used. Vector MV - REF and time interval tr - REF. It is also possible to directly use the motion vector of the image whose reference decoding order is followed. In this case, the motion vector MV45 and the time interval TR45 can be directly applied as the motion vector MV_REF and the time interval tr_rEF as in Equation 10. By 41 1258991 发明, invention description ———— This method's blocks belonging to the image P44 with reference to the motion vector can realize motion compensation by memorizing only the state of one of the two motion vectors, and thus can decode The capacity of the motion vector memory of the device is suppressed to be small. In addition, when the direct mode is used to determine the motion vector and the reference image system 5 is used to display the two motion vectors of the second frame image in the chronological order, the two motion vectors MV43 and the movement may be requested. The vector MV44 is set to "〇" to perform motion compensation. By this method, each block belonging to the image P44 to be referred to the motion vector can not first memorize the motion vector, so that the capacity of the motion vector memory portion of the decoding device can be suppressed small, and the motion vector can be omitted for 10 Processing. Moreover, in the present embodiment, the case where the distance between the images is used for the motion vector to be referred to is scaled, thereby calculating the motion vector used in the direct mode, which may also refer to the motion vector to be referred to. Calculate by constant multiple. Here, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of blocks of 15 blocks or a plurality of image units. (Embodiment 5) The present invention is not limited to the encoding method or the decoding method shown in the above-described Embodiments 1 to 4, and the encoding method or the decoding method can be realized using the motion vector calculating method shown below. 20 Figure 12 is a diagram illustrating the coded block or decoded block referenced in the direct mode for calculating the motion vector, with two motion vectors referenced to display the two frames in front of the time sequence. The schema of the action in the case. The image P23 is taken as an image of the object currently encoded or decoded. If the block to be encoded or decoded is set to block MB1, then it will be viewed 42 1258991

玖: The two motion vectors necessary for the SMI use the block MB2 located at the same position as the image P24 after the encoded or decoded reference image (the second reference image specified by the second reference index). It is determined by the motion vector. Moreover, in the 12th picture, the block MB1 is the processing target block, and the block MB1 and the block 5 MB2 are located at the same position on the image, and the motion vector MV21 and the motion vector MV22 are used to block the block MB2. The motion vectors used in encoding or decoding are referred to the image P21 and the image P22, respectively. Further, the image P21, the image P22, and the image P24 are encoded or decoded images. Further, the time interval TR21 indicates the time interval between the image P21 and the image P24, and the time interval 1022 indicates the time interval between the image P22 and the image P24, and the time interval TR21' indicates the image P21 and the image P23. The time interval between the time intervals TR24' represents the time interval between the image P23 and the image P24. The motion vector calculation method first encodes or decodes the previous motion vector (first motion vector) MV21 among the motion vectors of the block MB2 of the reference image 15 P24 as shown in Fig. 12, as for the movement of the block MB1. The vector MV21' and the motion vector MV24' are calculated by the following equation. MV2r = MV21 X TR21'/TR21 MV24, = - MV21 X TR24, /TR21 20 Two directions prediction is performed from the image P21 and the image P24 using the motion vector MV21' and the motion vector MV24'. Alternatively, instead of using only the motion vector MV21, the motion vector MV21' and the motion vector MV24 may be calculated, and instead, only the motion vector encoded or decoded in the motion vector of the block MB2 of the reference image P24 may be used. 2 moving vector) MV22 and calculating block 43 1258991 发明, invention describes the motion vector of MB1. Further, the motion vector of the block MB1 can be determined using both the motion vector MV21 and the motion vector MV22 as shown in the embodiment 1 to the embodiment 4. In the case of selecting any of the motion vector MV21 and the motion vector ~ MV22, any one of them can be selected. The 5 situation is to select the motion vector of the block that is encoded or decoded in time, ~ by the encoding device, Any one of the decoding means can be arbitrarily set in advance. Moreover, the image P21 can be compensated for movement in the short term Buffer or in the Long Term Buffer. The short-term memory and long-term memory will be described later. 10 Fig. 13 is a diagram for explaining the case where the calculated motion vector is calculated in the direct mode and the decoded block or the decoded block has a reference to the two motion vectors of the two frames in the rear in the display time sequence. The schema of the action below. The image 22 is taken as an image of the object currently encoded or decoded. If the block to be encoded or decoded is set to block ΜΒ1, then the two necessary motion vectors that are considered to be 15 are used after the encoding or decoding is completed (as specified by the second reference index). The image of the second reference image is determined by the motion vector of the block ΜΒ2 at the same position. Moreover, in the picture of Yudi 13, the block MB1 is the processing target block, and the block MB1 and the block 'MB2 are located at the same position on the image, and the motion vector mV24·20 and the motion vector MVM system will The motion vector used when the block MB2 is encoded or decoded is referred to the image P24 and the image P25, respectively. Further, the image P21, the ^ image P23, the image P24, and the image P25 are encoded or decoded images. Further, the time interval TR24 represents the time interval between the image P23 and the image P24, the time interval TR25 represents the time interval 44 1258991 33⁄4^ between the image P23 and the image p25, and the time interval TR24' represents the image P22 and the figure Like the time interval between P24, the time interval TR21' represents the time interval between the image P21 and the image P22. The motion vector calculation method uses only the motion vector MV24 of the input image P24 of the block MB2 of the reference image 5 P23 as shown in FIG. 13, and the motion vector MV21' and the motion vector MV24' of the block MB1 are as follows. Formulated to calculate. MV215 = -MV24 X TR2T/TR24 MV24, = - MV24 X TR24, /TR24 10 Two directions prediction is performed from the image P21 and the image P24 using the motion vector MV21' and the motion vector MV24'. Further, in the case where only the motion vector MV25 of the image P25 of the block MB2 of the input reference image P23 is used as shown in Fig. 14, the motion vector MV21' and the motion vector MV25' of the block MB1 are calculated according to the following equation. Further, the time interval TR24 represents the time interval between the image P23 and the image P24, the time interval TR25 represents the time interval between the image P23 and the image P25, and the time interval TR25' represents the image P22 and the image P25. The time interval between the time intervals TR21' represents the time interval between the image P21 and the image P22. 20 MV215 = -MV25 x TR2T/TR25 MV25' = MV25 X TR257TR25 Two directions prediction is performed from the image P21 and the image P24 using the motion vector MV21' and the motion vector MV25'. Figure 15 is a diagram for explaining the calculation of the motion vector in the direct mode and referring to the coded block or the decoded block of 45 1258991, with the reference to the image of one of the frames in front of the order of the coins. The schema of the action in the case of a motion vector. The image P23 is taken as an image of the object currently encoded or decoded. If the block to be encoded or decoded is set to block MB1, then it is considered as -5. The two necessary motion vectors are used after the coded or decoded image (after the second reference index). The image P24 of the designated second reference image is determined by the motion vector of the block MB2 at the same position. Further, in Fig. 15, the block MB1 is the processing target block, and the block MB1 and the block MB2 are located at the same position on the image, and the motion vector 10 MV21A and the motion vector MV21B are the block MB2. The forward direction motion vector is used for encoding or decoding, and the image P21 is referred to. Further, the image P21, the image P23, and the image P24 are images that have been encoded or decoded. Further, the time interval TR21A, the time interval TR21B represents the time interval between the image P21 and the image P24, the time interval TR21' represents the time interval between the image P21 and the image P23 of Fig. 15, and the time interval TR24' represents the image P23 Time interval from image P24. The motion vector calculation method uses only the forward motion vector 'MV21A of the input image P21 to be input to the block MB2 of the reference image P24 as shown in Fig. 15, as for the motion vector MV21A' of the block MB1, the motion vector·20 MV24' is calculated by the following formula.

MV21A' = one MV21A X TR21, /TR21A MV24, two-MV21A X TR24, /TR21A * Two directions prediction is performed from the image P21 and the image P24 using the motion vector MV21A' and the motion vector MV24'. 46 1258991

Further, the WMMM may calculate the motion vector of the block MB1 using only the forward motion vector MV21B of the image P21 of the block MB2 of the reference image P24. Further, it is also possible to determine the motion vector for the block MB1 using both the motion vector MV21A and the motion vector MV21B as shown in the embodiment 1 to the embodiment 4. In the case where any of the forward motion vector MV21A and the forward motion vector MV21B are selected, any one of the cases can be selected to select a motion vector that is encoded or decoded in time (previously described in the code column). The coding device and the decoding device are arbitrarily set. Here, the so-called motion vector that has been encoded or decoded in time means the 1st 10th motion vector. Further, the image P21 can be compensated for movement in a short term Buffer or a Long Term Buffer at any place. The short-term memory and long-term memory will be described later. Further, in the present embodiment, the case where the distance between the 15 images is used for the motion vector to be referred to is scaled, thereby calculating the motion vector used in the direct mode is described. The vector is calculated as a constant multiple. Here, the constant used in the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. Further, after calculating the right side of each calculation formula in the calculation formulas of the above-described motion vector MV21', motion vector MV24', shifting motion vector MV25', and motion vector MV21A', it is also possible to obtain the predetermined motion vector accuracy. As for the accuracy of the motion vector, there are 丨/2 pixels, 丨/3 pixels, and 1/4 pixel precision. And the accuracy of this motion vector is determined, for example, by a block unit, an image unit, and a sequence unit. 47 1258991 Now: Inventions®^—— (Implementation 6) This embodiment 6 uses the 16th to 18th drawings to illustrate the motion vector calculation method, which is used to determine the motion vector in the direct mode. In the case of having two first five-square motion vectors with reference to the two-frame image in front of the chronological order, a method of calculating the target motion vector by only one of the two forward motion vectors can be scaled. Further, the block MB1 is a processing target block, the block MB1 and the block MB2 are located at the same position, and the motion vector MV21 and the motion vector MV22 are motion vectors used when the block MB2 is encoded or decoded, and The image P21, 10 image P22 is referred to, respectively. Further, the image P21, the image P22, and the image P24 are images that have been encoded or decoded. Further, the time interval TR21 represents the time interval between the image P21 and the image P24, the time interval TR22 represents the time interval between the image P22 and the image P24, and the time interval TR21' represents between the image P21 and the image P23. The time interval TR22' represents the time interval between the image P22 15 and the image P23. In the first method, the block MB2 of the reference image P24 as shown in Fig. 16 has two forward direction motion vectors of the front side motion vector MV21 to be input image P21 and the front side motion vector MV22 to be input image P22. At the time of the object image P23, only the motion vector MV22 to which the 20 image P22 is displayed in order of display time is input is used, and the motion vector MV22' of the block MB1 is calculated by the following equation. MV22, two MV22 X TR22, /TR22 perform motion compensation from image P22 using motion vector MV22'. In the second method, as shown in Fig. 17, the block 1258901 of the image P24 is displayed ———--------------------------- ----------------- MB2 has the front side motion vector MV21 of the input image P21 and the two front direction motion vectors ' of the forward motion vector MV22 of the image P22 to be input. In the object image P23, only the motion vector MV21 of the image P21 to be input in order to display the time sequence is used, and the motion vector * 5 MV21' of the block MB1 is calculated by the following equation. ~ MV21, = MV21 X TR21, /TR21 ' Motion compensation is performed from the image P21 using the motion vector MV21'. By the first and second methods, the block MB2 belonging to the reference image P24 of the motion vector to be referred to can realize the motion compensation by first memorizing the state of 10 out of the two motion vectors. The capacity of the motion vector memory unit is suppressed to be small. Further, it is possible to perform motion compensation from the image P23 of the image in the vicinity of the display time sequence, similarly to the embodiment 1, by using the forward direction motion vector MV21. The motion vector MVN (not shown in Fig. 15) used at this time is calculated according to the following equation. MVN=MV21 X TR22VTR21 Further, in the third method, as shown in FIG. 18, the motion vector block MV21 and the motion vector MV22 are used, and the motion compensation block is obtained from the image P21 and the image P22, respectively. The average image is taken as a cropped image of ~20 in motion compensation. Although the third method increases the amount of calculation, it improves the accuracy of the movement compensation. Further, the motion vector MVN and the motion vector MV22 may be used, and the motion compensation block may be obtained from the image P22, and the average image 49 1258991 发明 and the invention description may be used as the interpolation image in the motion compensation. Moreover, in the present embodiment, the case where the distance between the images is used for the motion vector to be referred to is scaled, thereby calculating the motion vector used in the direct mode, which may also refer to the motion vector to be referred to. Calculate by 5 constant times. Here, the constant used in the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. Further, after calculating the equations of the motion vector MV21', the motion vector MV22', and the motion vector MVN, the right side of each equation can be calculated within the predetermined motion vector accuracy. As for the accuracy of the motion vector, there are 1/2 10 pixels, 1/3 pixels, and 1/4 pixel precision. And the accuracy of this motion vector is determined, for example, by a block unit, an image unit, and a sequence unit. The above embodiment 6 illustrates a reference image used for determining a motion vector for encoding or decoding a target block in a direct mode, and has a reference to two forward motion vectors of two frames of images in front of the display time sequence, !5, the same applies to the two rearward movement vectors (the second motion vector of the reference image specified by the second reference index) in the rear display of two frames of images, and only two rear motion vectors can be moved. One of them is scaled to calculate the object motion vector. The following is a description of FIG. 19 to FIG. 22, the block MB1 is a processing target block, the block MB1 and the block MB2 are located at the same position of the phase 2, and the motion vector MV24 and the motion vector MV25 are the area. The motion vector used when the block MB2 is encoded or decoded (the second motion vector of the reference image is specified by the second reference index). Further, the image P21, the image P23, the image P24, and the image P25 are images that have been encoded or decoded. Also, the time interval TR24 represents the time between the image P23 and the image P24 50 1258991 ------------------------------ -------------- Interval, time interval TRM represents the time interval between image P23 and image P25, time interval TR24' represents the time interval between image P22 and image P24 The time interval TR25' represents the time interval between the image P22 and the image P25. 5 As the first method, the block MB2 of the reference image P23 as shown in Fig. 19 has two rearward direction movements of the motion vector MV24 after the input image P24 and the motion vector MV25 after the input image P25. In the case of the vector, the motion vector MV24 of the image P24 to be input in order of display time is used only for the object image P22, and the motion vector 10 MV24' of the block MB1 is calculated by the following equation. MV24, = MV24 X TR24, /TR24 Motion compensation from image P24 using motion vector MV24'. Further, it is also possible to perform motion compensation from the image 15 P23 of the image in the vicinity of the display time sequence, similarly to the embodiment 1, using the forward direction motion vector MV24. The motion vector MVN1 (not shown) used at this time is calculated according to the following equation. MVN1-MV24 X TRN1/TR24 In the second method, the block MB2 of the reference image PM as shown in Fig. 20 has a moving vector MV24 to be input after the image P24, and a moving image 205 after being input into the image P25. When the vector MV25 moves the vector in the two backward directions, the object image is only used to move the vector MV25 in the backward direction of the image P25 to be displayed in the chronological order, and the motion vector MV25' of the block MB1 is expressed as follows. Calculation. MV25'2 MV25 X TR25'/TR2 1258991 发明y Description of the Invention Motion compensation is performed from the image Ρ25 using the motion vector MV25'. By the first and second methods, the block ΜΒ2 belonging to the reference image Ρ23 of the motion vector to be referred to can perform motion compensation by first memorizing the state of one of the two motion vectors, so that the motion vector can be memorized. The volume of the part is suppressed to a small extent. Further, it is also possible to perform motion compensation from the image f image P23 of the image in the vicinity of the display time sequence, similarly to the embodiment 1, using the backward direction motion vector MV25. The motion vector MVN2 (not shown) used at this time is calculated according to the following equation. 10 MVN2-MV25 x TRN1/TR25 Further, as the third method, as shown in FIG. 21, the motion compensation area is obtained from the image P24 and the image P25 using the motion vector MV24' and the motion vector MV25' obtained as described above. The block is treated as a patch image in motion compensation. 15 Although the third method increases the amount of calculation, it improves the accuracy of the object image P22. Furthermore, the motion compensation block can also be obtained from the image P24 using the above-described motion vector MVN1 and motion vector MVN2, and the average image thereof can be used as the interpolation image in the motion compensation. Further, as shown in FIG. 22, in the case where the reference image used for determining the motion vector in the direct mode has a reference to display the time sequence to move the vector behind one of the rear frame images, for example, To calculate the motion vector MV24'. Motion compensation is performed from the image P24 using the motion vector MV 24'. 52 1258991 wvwmmm Further, it is possible to perform motion compensation from the image - P23 of the image in the vicinity of the display time sequence, similarly to the implementation of the state 1, by using the backward direction vector MV25. The motion vector MVN3 (the pattern is not 'displayed) used at this time is calculated according to the following equation.辠5 MVN3-MV24 x TRN1/TR24 - Further, in the present embodiment, the use of the 19th to 22nd drawings illustrates the case where the vector has two reference directions in the rear direction of the two frames in the display time sequence. In the case where there is a reference shifting vector in the rear direction of one of the φ frame images in the display chronological order, the rear direction is shifted by 10 vectors to be scaled to calculate the object motion vector, but this may not The object motion vector is calculated by referring to the motion vector of the peripheral block in the same image by using the rear motion vector, and the image motion vector can also be calculated by referring to the motion vector of the peripheral block in the same image. First, the first calculation method will be described. Fig. 23 shows the positional relationship between the moving 15 vector and the target block to be referred to at this time. The block MB1 is a target block, and is a moving direction of a block which refers to three pixels included in the positional relationship of A, B, and C. However, in the case where the position of the pixel C is out of the picture or in a state where the encoding/decoding is not completed, the motion vector including the block of the pixel D is used instead of the block including the pixel C. The motion vector actually used in the direct mode is set to obtain the state of the center 移动 of the motion vector included in the three blocks of the A, B, and C pixels to which the reference is made. . In order to obtain the state of the center 移动 of the motion vector of the three blocks, it is not necessary to describe in the code column why the additional information of the motion vector is selected among the three motion vectors, and the neighboring block can be obtained. The actual 53 1258991 MBT invention of MB1 describes the movement vector of the moving movement. In this case, only the forward reference (reference to the first reference image) can be used for motion compensation using the determined motion vector, and the determined motion vector and the parallel motion vector can be used to refer to the two directions. (Reference to the first reference image and the second reference image) Five lines of motion compensation are performed. Next, the second calculation method will be described. The second calculation method does not acquire the center 値 as in the first calculation method, and obtains the motion vector having the highest coding efficiency from the motion vectors included in the three blocks including the A, Β, and C pixels to be referred to. The motion vector used in the actual 10 direct mode. In this case, only the forward reference (reference to the first reference image) can be used for motion compensation using the determined motion vector, and the determined motion vector and the parallel motion vector can be used to refer to the two directions. (Reference to the first reference image and the second reference image) Motion compensation is performed. The information indicating the motion vector having the highest coding efficiency is, for example, shown in Fig. 24(a), and shows the information of the direct mode output from the mode selection unit 107, and is added to the block in the code sequence generated by the code sequence generation unit 103. The front field. Further, as shown in Fig. 24(b), information indicating the vector with the highest coding efficiency can be added to the front field of the macroblock. Further, the information indicating the vector having the highest coding efficiency is, for example, a number identifying a block including 2 pixels to be referred to, and an identification number attached to each block. Further, when the block is identified by the identification number, only one of the identification numbers attached to each block may be used, or only one of the motion vectors used when the block corresponding to one identification number is encoded may be used. The most efficient coding vector can also be represented by the majority of the motion vector when the motion vector is a majority. 54 1258991 玖— ------------------------- – the motion vector with the highest coding efficiency or each of the two motion vectors (references to the first reference picture and the second reference picture) is attached to the identification number of each block. The most efficient moving vector. By using such a moving vector selection method, it is possible to obtain a motion vector with the highest coding efficiency. However, since it is necessary to describe the additional information indicating which motion vector is selected in the code sequence, it is necessary to additionally use the above code amount. Further, the third calculation method will be described. The third calculation method sets the motion vector with the reference index 参照 of the reference image of the reference vector as the motion vector used in the direct mode. The so-called reference index is the smallest, generally referring to an image that is close in order of display time. Or the most efficient moving vector. Therefore, by using such a moving vector selection method, the moving vector used in the direct mode is generated in the display chronological order, or the motion vector of the image with the highest encoding efficiency is used, and the encoding efficiency can be improved. 15 In the case where three reference vectors are identical among the three motion vectors, the center of the three motion vectors can be obtained. In the case where the movement vectors of the reference images whose reference index 値 is the smallest among the three movement vectors are two, for example, one of the two movement vectors is set to be fixedly selected. In the case of using the 23rd picture as an example, in the motion vector including the three blocks of 20 A, B, and C, the reference index of the two blocks including Pixel A and P1 is the smallest and In the case of referring to the same reference image, it is only necessary to obtain the motion vector of the block including the pixel A. However, in the motion vector of each of the three blocks including the pixel A, the pixel B, and the pixel C, the pixel A and the pixel C are included in the 1258999 ——— ---------- ----------------------------------- The reference index of the two blocks is the smallest and refers to the same reference image. In this case, it is preferable to obtain a motion vector including a block of the pixel A having the closest positional relationship to the block BL1. Further, the center ridge may be provided so that the center 値 is obtained for each of the horizontal direction component and the vertical direction component, and the size (absolute 値) of each motion vector is set to obtain the center 値. Further, in the case where the center of the motion vector is as shown in FIG. 25, the reference image at the rear is at the same position as the block BL1, and the block including the pixel A, the pixel B, and the pixel C, respectively, and The block of 10 pixels D as shown in Fig. 25 is included to be set to obtain the center 値 of the motion vectors of the five blocks. When the reference image near the encoding target pixel is used and the reference image at the rear is located at the same position as the block BL1, since the number of blocks is set to an odd number, once the block including the pixel D is used, the calculation of the motion vector is simply made. Central 値 processing. Further, in the case where the rear reference picture covers a plurality of blocks in the same area as the 15 block BL1, the motion vector in the block of the largest area overlapping the block BL1 among the plurality of blocks can be used. The motion compensation of the block BL1 is performed, or the majority of the block areas in the reference image corresponding to the block BL1 can be distinguished, and the block BL1 is moved and compensated for each of the divided blocks. 20 Also, explain with a specific example. As shown in FIG. 26 or FIG. 27, when all the blocks including the pixel A, the pixel B, and the pixel C are referred to the motion vector of the image that is further ahead than the image to be encoded, the above-described first calculation method can be used. To any of the methods of the 3 g method. 1258991 ----------------------------------------------- - the same When all the blocks including the pixel A, the pixel B, and the pixel C are referred to the motion vector of the image further than the encoding target image, any one of the above-described first to third calculation methods can be used. - The second is to show the pattern shown in Figure 3. The third diagram shows a case where all of the blocks including each of the five pixels A, B, and C have a motion vector that refers to an image that is more forward and rearward than the image to be encoded. According to the first calculation method described above, the motion vector used before the motion compensation of the block 肇BL1 is selected by the motion vector MVAf, the motion vector MVBf, and the motion vector MVCf, by the motion vector MVAb 10 and the motion vector MVBb. The center of the motion vector MVCb is selected to use the motion vector after the motion compensation of the block BL1. Moreover, the motion vector MVAf is a forward motion vector of a block including the pixel a, the motion vector MVAb is a rear motion vector of the block including the pixel A, and the motion vector MVBf is a forward motion vector of the block including the pixel B, and the motion vector 15 MVBb is the rear motion vector of the block containing pixel B, the motion vector MVCf is the forward motion vector of the block containing pixel C, and the motion vector MVCb is the rear motion vector of the block containing pixel C. Further, the moving vector MVAf or the like is not limited to the image shown in the drawings. This case is also the same in the description of <. * 20 According to the second calculation method described above, the motion vector with the highest coding efficiency is obtained from the motion vector referenced by the motion vector MVAf, the motion vector MVBf, and the motion vector MVCf, and the motion vector MVAb, the motion vector MVBb, and the motion vector are obtained. The moving vector with the highest coding efficiency is obtained from the motion vector referenced by the MVCb, and the motion vector is actually used in the direct mode to make 57 1258991 玖 = invention description. In this case, the motion vector with the highest coding efficiency may be obtained from the motion vector referenced by the motion vector MVAf, the motion vector MVBf, and the motion vector MVCf, and the motion compensation may be performed only by the forward reference, and the determined motion may be used. The motion vector and the parallel motion vector 5 are motion compensated in a two-way reference. It is also possible to maximize the coding efficiency without separately selecting the motion vectors of the forward reference and the backward reference to select a block and use the block to perform motion compensation with the motion vectors of the previous reference and the backward reference. In this case, it is compared with a block indicating that a pixel having a motion vector with the highest coding efficiency is selected, and a block having a pixel having a motion vector with the highest coding efficiency as the back is selected. In the case, since the information indicating the selection is small, the coding efficiency can be improved. Further, selecting one of the blocks may be (1) a block including pixels having a reference index 値 having a reference index 图像 of an image to which the forward referenced motion vector is the smallest, and (2) having each pixel having The reference index of the image referenced by the motion vector referenced by the block before the 15th block is added to the reference index of the image to which the motion vector referenced by the back reference is added, and the block which is added to the minimum is obtained, and (3) is obtained. The center of the reference index of the image referenced by the motion vector referred to in the forward direction is set to include the block having the pixel of the motion vector with the center 値 referenced before, and the 20-movement vector of the rear reference is set to this area. The block has a motion vector to be referred to by the latter, and (4) obtains a center 値 of the reference index of the image referred to by the motion vector referred to later, and is provided to include a pixel having a motion vector with a reference to the center 値. The block, and the moving vector of the front reference may be set to any of the above methods, such as the moving vector referred to by the previous block. Again, after 1258991

When the moving vector referenced by the SWJ side refers to all the same images, the selection method of the blocks of the above (1) and (3) can be applied. In the third calculation method, the motion vector MVAf, the motion vector MVBf, and the motion vector referenced by the motion vector referenced by the motion vector referenced by the motion vector MVCf are set to be the minimum motion vector before the direct mode is used. (first reference) motion vector. Or the motion vector MVAb, the motion vector MVBb, and the motion vector referenced by the motion vector referenced by the motion vector after the motion vector MVCb are set to the minimum motion vector, which is set to be used in the direct mode (second reference). Move the vector 10 . Further, in the third calculation method, the motion vector referred to before the reference index of the reference image is minimized is the motion vector referred to by the block BL1, and the reference index of the reference image is minimized. The motion vector is set to the motion vector referenced after the block BL1, however, the motion vector of the reference image 値 may be used to derive the two motion vectors of the block BL1 using any one of the first or the least of the reference index ,, and Block BL1 is motion compensated using the derived motion vector. Next, the situation shown in Fig. 31 will be explained. Fig. 31 shows that the pixel A has a motion vector for each of the front and rear images, and the pixel B has only the motion vector of the image with reference to the front, and the pixel C has only the motion vector with reference to the rear 20 images. When there is a block including a pixel having only a motion vector of a reference image, the motion vector of the other image of the reference block is set to 0, and the calculation method of the above FIG. 30 is used for motion compensation. can. Specifically, using the first calculation method or the third calculation method in Fig. 30, it is assumed that 1258991 玖, 葰明明—for MVCf=MVBb=0. That is, when the first calculation method calculates the motion vector before the block BL1, the pixel C refers to the motion vector MVCf of the front image as MVCf=0, and calculates the center of the motion vector MVAf, the motion vector MVBf, and the motion vector MVCf. . Further, when the vector is moved after the block BL1 in the calculation area, the pixel B refers to the motion vector MVBb of the front picture as MVBb = 0, and the center 値 of the motion vector MVAb, the motion vector MVBb, and the motion vector MVCb is calculated. In the third calculation method, the pixel C refers to the motion vector MVCf of the front image and the motion vector MVBb of the pixel B with reference to the rear image is set to 10 MVCf=MVBb=0, and the reference image to which the motion vector of the block BL1 refers is calculated. The reference index 値 is the smallest motion vector. For example, when the image including the pixel A referring to the first reference index "0" includes the image in which the pixel B refers to the first reference index "1", the smallest first reference index is only "〇". Since the image having the smallest first reference index is referred to only by the movement 15 vector MVBf of the image in front of the block including the pixel B, the motion vector MVBf is set as the forward motion vector of the block BL1. Further, for example, when any one of the pixel A and the pixel C refers to the second reference index, for example, when the second reference index is a "0" image, the pixel B is referred to the rear image. The motion vector MVBb is set to 20 MVBb=0 to calculate the center 値 of the motion vector MVAb, the motion vector MVBb, and the motion vector MVCb. The motion vector obtained as a result of the calculation is set as the square motion vector after the block BL1. Next, the situation shown in Fig. 32 will be explained. Figure 32 shows that the pixel A has a motion vector for each of the front and rear images. The pixel B has only 60 1258991 玖. The invention describes a motion vector with reference to the image in front. The pixel C does not have a motion vector and is displayed in the screen. The case of coding. As described above, when the block including the pixel C of the reference object is intra-coded, the motion vector of the image before and after the reference block is set to 5 and is used for motion compensation. The calculation method of the 30 figure can be used. Specifically, it can be calculated by setting MVCf and MVCb=0. Further, the case of Fig. 30 is MVBb = 0. Finally, the situation shown in Fig. 33 will be explained. Fig. 33 shows a case where the pixel C is encoded in the direct mode. 10 When the pixel of the reference object has a block coded in the direct mode, the motion vector used when the block coded by the direct mode is encoded is used, and the motion compensation of the block BL1 is performed by using the calculation method of FIG. can. Further, whether the motion vector is the front reference or the backward reference is determined based on the time information of the image to be referred to and the image of the coded image. Therefore, in the case of distinguishing between the front reference and the backward reference, the case where the motion vector is derived is determined by the motion vector of each block based on the time information of each image to determine which of the forward reference and the backward reference. . 20 Also an example of a combination of the above-described calculation methods will be described. Figure 34 shows the order of the motion vectors determined by the direct mode. Fig. 34 is an example of a method of determining a motion vector using a reference index. Further, RidxO and Ridx1 shown in Fig. 34 are the reference indexes described above. Fig. 34(a) shows the procedure for determining the motion vector based on the first reference index RidxO, and Fig. 34(b) 1258991 m-- indicates the order in which the motion vector is determined by the second reference index Ridx1. First, explain Figure 34 (4). In step S37〇1, the number of blocks in which the reference image is calculated using the first reference index 5 RidxO among the block including the pixel A, the block including the pixel b, and the block including the pixel C is calculated. When the number of blocks calculated in step S3701 is "〇", the number of blocks of the reference image is calculated using the second reference index Rixx1 in step S37〇2. When the number of blocks calculated in step S3702 is "1", the motion vector of the coding target block is set to "0" in step S3703, and the coded 10 object block is motion-compensated in two directions. On the other hand, if the number of blocks calculated in step S3704 is "1" or more, the motion vector of the coding target block is determined by the number of blocks in which the second reference index Ridx1 exists in S3704. For example, the motion compensation of the coding target block is performed using the motion vector determined by the number of blocks in which the second reference index Ridx1 exists. When the number of blocks calculated in step S3701 is "丨", the movement vector of the block in which the first reference index RidxO exists is used in step S37〇5, and the number of blocks calculated in step S3701 is "2". In step S37〇6, the block in which the first reference index Ridx0 does not exist is temporarily set to have a motion vector of 〇 in the 20th first reference index Ridx0, and a motion vector corresponding to the center 三 of the three motion vectors is used. When the number of blocks calculated in step S3701 is "3", a motion vector corresponding to the center 三 of the three motion vectors is used in step S3707. Alternatively, the motion compensation in step S3704 may be set to 1258991 - a motion vector in two directions using a motion vector. The motion compensation in the two directions may also be performed by a motion vector and a motion vector of the same direction and a motion vector of the opposite direction, for example, when the motion vector is scaled, or may be used. The motion vector 5 in one direction in which the motion vector is the same is performed using the motion vector in which the motion vector is "〇". Next, explain Figure 34 (b). In step S3711, the number of blocks in which the second reference index Ridx1 exists is calculated. If the number of blocks calculated in step S3701 is "3", the number of blocks in which the first reference index RidxO exists is calculated in step S3712. When the number of blocks calculated in step S3712 is "0", the motion vector of the coded object block is set to "0" in step S3713, and the coding target block is motion-compensated in two directions. On the other hand, if the number of blocks calculated in step S3712 is "1" or more, the motion vector of the coding target block is determined in S3714 by the number of blocks of the first reference index RidxO. For example, the motion compensation of the coding target block is performed using the motion vector determined by the number of blocks in which the first reference index RidxO exists. When the number of blocks calculated in step S3711 is "1", the motion vector 20 of the block in which the second reference index Ridx1 exists in step S3715 is "2" in the case where the number of blocks in the step S3711 is "2". In step S3716, the block in which the second reference index Ridx1 is not present is temporarily set as the motion vector included in the second reference index Ridx1, and the motion vector corresponding to the center 三 of the three motion vectors is used. 63 1258991 疚 '— 任 丽 — If the number of blocks calculated in step S3711 is "3", a motion vector corresponding to the center 三 of the three motion vectors is used in step S3717. Further, the motion compensation in step S3714 may be set to a motion vector in two directions using a motion vector. The motion compensation in the two directions can also be performed by a motion vector and a motion vector in the same direction and a motion vector in the opposite direction, for example, when the motion vector is scaled, or can be used. A motion vector in the same direction as a motion vector is performed using a motion vector in which the motion vector is "〇". Further, although the above description of Fig. 34 (a) and Fig. 34 (b) has been described above, it is possible to use 10 for both the processing and only one of the processing. However, in the case of using one of the processes, for example, when the process from step S3701 shown in (a) of FIG. 34 is performed, and the process of step S3704 is further performed, the case shown in FIG. 34(b) can be performed. The processing of S3711 is as follows. Further, in the case of the processing of S3704, the processing below step S3712 is not performed in the processing below step S3711, so that the motion vector can be determined intentionally. Further, in the case of using both of the processing of Fig. 34 (a) and Fig. 34 (b), either one of the processes may be performed first, or may be carried out in combination. In addition, when the block around the coding target block is the coded block by the direct mode, the image of the motion vector used when the block coded by the direct mode is encoded is used. The reference index is set to the block coded by the direct mode and is the one of the blocks around the coding target block. The method of determining the motion vector will be described below by taking an example of a specific block of sex. Fig. 35 shows the type of the motion vector which each block of the coding target block BL1 has. Figure 35 (a) has a pixel block of 64 1258991 ----------------------------------- ------------- Intra-coded block, block with pixel B has a motion vector' is a block that is motion compensated with one motion vector, block with pixel c A block' block with two motion vectors and motion compensation in two directions. Further, the block having the pixel B has a motion vector indicated by the second reference index Ridx1. The block having the pixel A is a block coded in the picture, so that it does not have a motion vector, that is, does not have a reference index. The number of blocks in which the first reference index Ridx is present is calculated in step S3701. As shown in FIG. 35, the number of blocks in which the second reference index Ridx is present is two springs. Therefore, in step S3706, the block 10 in which the i-th reference index Ridx is not present is temporarily set to have the MV= in the first reference index RidxO. A motion vector of 0 is used, and a motion vector equivalent to the center 三 of the three motion vectors is used. It is possible to use the motion vector to compensate the motion of the coding target block in two directions or to use the second reference index Ridx1 shown below to use other motion vectors for motion compensation in two directions. 15 The number of blocks in which the second reference index Ridx1 exists is calculated in step S7311. As shown in Fig. 35, the number of blocks in which the second reference index Ridx1 exists is i-corrected. Therefore, the motion vector of the block in which the second reference index Ridx1 exists is used in steps S3 to l5. 'Another example of combining the above-described calculation methods will be described. The 36th '2' diagram shows the order of the motion vectors of the coding target block based on the reference index 图像 of the image to which the motion vectors of the blocks A, B, and C respectively have reference. (a) and (b) of FIG. 36 show the order in which the motion vector is determined based on the first reference index RidxO, and FIG. 36(c) shows that the order of the motion vector is determined based on the second reference index Ridx1. 65 1258991 ------------------------ Also, Fig. 36 (a) shows the case of the order based on the first reference index Ridx(). 36(c) shows the order based on the second reference index Ridx1, and FIG. 36(b) shows the order based on the first reference index Ridx〇. FIG. 36(d) shows the second Since the reference index Ridx1 is in the order of the reference 5, the following description will be described only with reference to Fig. 36 (a) and Fig. 36 (b). First, the figure is shown in Fig. 36 (a). It is judged that it is valid in step S3801. Whether or not a minimum first reference index Ridx0 can be selected among the first reference index Ridx0. In step S3801, it is determined that 10 of the valid first reference index Ridx0 can select a minimum first reference index RidxO, in the step S3802 uses the selected motion vector. In step S3801, in the case where the smallest first reference index RidxO among the valid first reference index Ridx0 is a majority In step S38〇3, the motion vector of the block selected according to the priority order is used. Here, the priority order is the block having the pixel A, the block having the pixel B, and the pixel C. The order of the blocks determines the motion vector used for the motion compensation of the coding target block. In the case where there is no valid first reference index RidxO in step S3801, a process different from S3802 or S3803 is performed in step S3804. Fig. 34(b) shows the processing of step S3711 and below. Next, it is explained in Fig. 36(b). The difference between Fig. 36(b) and Fig. 36(a) is based on Fig. 36 ( The processing in step S38〇3 and step S3804 in a) is set to the point of step S3813 shown in Fig. 36(b). In step S3811, it is judged whether or not the first reference index 66 is valid.

Among the RidxO, a minimum first reference index Ridx0 can be selected. In the case where a minimum first reference index Rixx0 can be selected among the valid first reference indices Ridx0 in step S3811, the selected motion vector is used in step S3812. When there is no valid first reference index RidxO in step S3811, a process different from S3812 is performed in step S3813. For example, the following processing may be performed in step S3711 described in Fig. 34 (b). Further, the first reference index Ridx which is valid as described above refers to the first reference index Ridx0 which is denoted as "〇" in Fig. 35(b), and is a reference index indicated by the movement amount. Further, the place marked as "X" in Fig. 35(b) indicates that the reference index is not assigned. Further, in step S38M in Fig. 36 (c) and step S3833 in Fig. 36 (d), the processing in step S3701 described in Fig. 34 (a) may be performed. The method of determining the motion vector will be described in detail below using the example of the specific block in Fig. 35. It is determined in step S3801 whether or not a minimum first reference index RidxO can be selected among the valid first reference indices RidxO. In the case shown in Fig. 35, the number of valid first reference indices RidxO is two, and in step S3801, it is determined that one of the valid first reference indices Ridx0 can select one of the smallest first reference indices RidxO. Step S38〇2 uses the selected motion vector. In the case where the smallest first reference index Ridx0 among the valid first reference indices RidxO is a large number in step S3801, the motion vector of the block selected by the priority order is used in step S3803. Here, the reference priority is to determine the motion vector used for the motion compensation of the coding target block in the order of the block having the pixel A, the block having the pixel B, and the block having the pixel C. In the case where the block having the pixel B has the same first reference index Ridx0 as the block having the pixel C, the first reference index Ridx0 in the block having the pixel B is used according to the 5 priority order, and the corresponding reference has The motion vector of the coding target block BL1 is performed by the motion vector of the first reference index Ridx0 in the block of the pixel B. At this time, the coding target block BL1 can be motion-compensated in only two directions using the determined motion vector, and the second reference cable 10 is used to refer to Ridx1 and the other motion vectors are used to perform the two directions as described below. The motion compensation 判断 determines in step S3821 whether a minimum second reference index Rixx1 can be selected among the valid second reference indices Ridx1. In the case shown in Fig. 35, since the effective second reference index RidxO 15 is one, the motion vector corresponding to the second reference index Rixx1 in the block having the pixel C is used in step S3822. Further, in the case where the block having no reference index is used, it is assumed that the motion vector having the magnitude of the motion vector is "〇", and the point of obtaining the total center of the motion vectors is set to have the size of the motion vector. The motion vector of "0" of 20 may be set to obtain the central 値 of the total of three motion vectors, or may be set to obtain the average 値 of the motion vectors of the block having the reference index. Moreover, the priority order of the above description may be set to, for example, a block having a pixel 、, a block having a pixel 、, and a block having a pixel C, and the order is 68 125899 疚 —, WMW — — The motion vector of the motion compensation for the coding object block. In this way, the motion vector used when the coding target block is motion-compensated is determined by using the reference index, whereby the motion-vector can be determined intentionally. Moreover, according to the above example, the coding efficiency can be improved. Further, since it is necessary to determine whether the motion vector is forward reference or backward reference without using the time information, the processing for determining the motion vector can be simplified. Further, considering the prediction mode of each block, the motion vector used for the motion compensation, and the like, although there are many patterns, it can be handled by the above-described series of processes. 10 In the present embodiment, the case where the motion vector used in the direct mode is calculated by using the distance between the image and the image for the reference motion vector is used, and the motion vector to be referred to can also be referred to. Constant multiple calculation. Here, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. 15 Also, the calculation method of the motion vector using the reference index Ridx〇 and Ridxl is not only the calculation method using the central unit but also the other calculation method groups. For example, in the third calculation method described above, in the block including the pixel A, the pixel B, and the pixel C, when the movement of the image having the smallest reference index is referred to and the motion vector is a majority, the calculation is not necessary. The movement of the block and the center of the heavy weight can also calculate the central 値 and the obtained movement and the tomb g are subjected to the motion vector used as the direct mode of the block BL1. Or, for example, select one of the most efficient moving vectors from among the smallest moving vectors whose index is the smallest. In addition, the direction vector of the block BL1 can be independently calculated from the movement direction of the backward direction and the amount of the movement of the block BL1 to 69 1258991. For example, the front direction motion vector and the backward direction motion vector can be calculated from the same motion vector. Further, any one of the forward direction motion vector and the backward motion vector obtained as a result of the calculation can be set as the motion vector of the block BL1. 5 (Embodiment 8) The reference block MB of the reference image of the present embodiment has a front direction (1st) motion vector which is referred to as a reference image stored in the long-term memory and is referred to as a first reference image. And a forward direction (second) motion vector that is referred to as a second reference image in which the reference image stored in the short-term memory is referred to. 10 Figure 37 shows the prediction of two directions in the direct mode in the case where only one reference image is stored in the long-term memory. The difference between the embodiment 8 and most of the embodiments described above is referred to the block MB2 front direction (1st) motion vector MV21 of the image 1 with reference to the reference image stored in the long-term memory. 15 The short-term memory system is used to temporarily store the memory of the reference image. For example, the image is stored in the order in which the image is stored in the memory (i.e., the order of encoding or decoding). When the image is newly stored in the short-term memory, the memory capacity is insufficient. The Japanese temple is deleted from the order of the image stored in the memory. Long-term memory does not necessarily limit the preservation of images in the order of time as in short-term memory. For example, the order in which the images are saved may correspond to the order of the images, or the order in which the addresses of the memory of the images are stored. Therefore, the motion vector M21 referring to the image stored in the long-term memory cannot be scaled according to the time interval. Long-term memory is not used for temporary storage as a short-term memory. 70 1258991 __________________________________________________________________-----------------------........ ...—......—--------------------------------------- ------------------------------------------- 玖, invention instructions photo And the person who continuously saves the reference image. Thus, the time interval corresponding to the motion vector stored in the long-term memory is much larger than the time interval corresponding to the motion vector stored in the short-term memory. In Figure 37, the boundary between long-term memory and short-term memory is shown in Figure 5 as a vertical dotted line. By this, information about the left image is stored in long-term memory, thereby Information is stored in short-term memory. The block MB1 of the image P23 is an object block. Further, the block MB2 is a reference image at the same position as the block MB1 in the reference image P24. Among the motion vectors of the block MB2 of the reference image P24, the first direction (first) movement to the tenth amount MV21 is the first motion vector referred to as the first reference image in the image P21 stored in the long-term memory. The rear direction (second) motion vector MV25 is a second motion vector that is referred to as the second reference image by the image P25 stored in the short-term memory. As before, the time interval TR21 of the image P21 and the image P24 should refer to the motion vector MV21 of the image stored in the long-term memory, and the time interval TR2 of the image P24 and the image P25 should be stored in the short-term memory. The motion vector MV25 of the image of the volume has a time interval TR21 of the image P21 and the image P24 which is much larger or indefinite than the time interval TR2 of the image P24 and the image P25. In this case, the present embodiment does not scale the motion vector of the block MB2 of the reference image P24 to determine the motion vector of the block MB1 of the object image P23, as in the embodiment described so far. The motion vector of the block MB1 of the object image P23 is calculated in the following manner. MV21 = MV21, 71 1258991 § MV24, = 0 The above expression indicates that the first motion vector MV21 stored in the long-term memory is originally regarded as the target image among the motion vectors of the block MB2 of the reference image P24. The first motion vector MV21'. The following equation indicates that the second motion vector MV24' to be stored in the block MB1 of the target image P23 stored in the short-term memory image P24 is much smaller than the first motion vector MV21', so that it is not necessary to pay attention. The second motion vector MV24' is treated as "0". As described above, the reference block MB has one motion vector referred to as the first reference image stored in the reference image of the long-term memory, and the reference image stored in the short-term memory as the second reference image. In the case of referring to one of the motion vectors, among the motion vectors of the block of the reference image, the motion vector stored in the long-term memory is used locally as the motion vector of the block of the target image and two are performed. Direction prediction. Further, if the reference image stored in the long-term memory is any of the first reference image or the second reference image, the motion vector MV21 of the reference image stored in the long-term memory may be referred to as the rear direction. Move the vector. Further, the second reference image is stored in the long-term memory, and when the first reference image is stored in the short-term memory, the scaling is applied to the motion vector of the reference image by referring to the 20 motion vector of the first reference image. . In this way, the processing of the two directions can be performed without using a relatively large or indefinite time of the long-term memory. Further, instead of using the motion vector to be referred to locally, the motion vector can be multiplied by two times. 72 1258991 P: 7 Description of the Invention Further, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. (Implementation 9) This embodiment shows that the reference block MB of the reference image has two directions prediction in the direct mode in the case of referring to the two forward motion vectors of the reference image stored in the long-term memory. . Fig. 38 is a view showing the two-direction prediction in the direct mode in the case where the reference block MB has two reference forward motion vectors stored in the reference image of the long-term memory. The difference between the implementation mode 9 and the implementation mode 8 is that both the motion vector MV21 and the motion vector MV22 of the block MB2 of the reference image refer to the image stored in the long-term memory. In Figure 38, the boundaries between long-term memory and short-term memory are shown as vertical dotted lines, whereby information about the left image is stored in long-term memory, whereby the image on the right The information is stored in short-term memory. The motion vector MV21 and the motion vector MV22 of the block MB2 of the image P24 refer to the image stored in the long-term memory. The motion vector MV21 corresponds to the reference picture P21, and the motion vector MV22 corresponds to the reference picture P22. 2〇 corresponds to the motion vector MV22 stored in the image P22 of the long-term memory, and the time interval TR22 between the image P22 and the image P24 is larger than the time interval TR25 of the image P24 and the image P25 stored in the short-term memory. Many, or uncertain situations. In Fig. 38, the image P22 corresponding to the motion vector MV22 and the image P21 corresponding to the motion shift vector MV21 are sequentially assigned, and the image P21 and the image P22 are stored in the long-term memory. The motion vector of the block MB1 of the object image is calculated as shown in Fig. 38 as follows. MV22, = MV22 5 MV24, = 0 The above expression indicates that among the motion vectors of the block MB2 of the reference image P24, the motion vector MV22 of the image P21 with the smallest reference allocation order is originally regarded as the object image P23. The movement vector MV22 of the block Mbl. The following equation indicates that the direction motion vector MV24' is smaller than the first motion vector MV21' after the block MB1 stored in the object image P23 of the short-term memory, so that it is not necessary to pay attention. The rear movement vector MV24' is treated as "0". As described above, the motion vector of the image having the smallest allocation order among the motion vectors stored in the block 15 of the reference image stored in the long-term memory is originally used locally, and is used as the motion vector of the block of the target image. It is possible to perform processing in two directions without using a relatively large amount of long-term memory or an indefinite time. Alternatively, the motion vector to be referred to may be used locally, and the motion vector may be multiplied by two times to perform prediction in two directions. Further, the constant used for the constant multiple may be changed in the case of encoding or decoding in a plurality of block units or a plurality of image units. Further, in the case where both the motion vector MV21 and the motion vector MV22 of the reference image block MB2 are referred to the image stored in the long-term memory, 74 1258991 发明, the invention says W- can also be selected to refer to the first reference map. Like moving the vector. For example, in the case where the MV 21 refers to the motion vector of the first reference picture and the MV 22 refers to the motion vector of the second reference picture, the motion vector of the block MB1 uses the motion vector MV21 for the picture P21 and the movement for the picture P24. Vector "5 0,, ° (Implementation Mode 10) This embodiment describes a motion vector calculation method in the direct mode described in Embodiment 5 to Embodiment 9. This motion vector calculation method is applied to image coding. It can be applied at the time of decoding. Here, the object block of the encoding or decoding 10 is referred to as an object block. Further, the block at the same position as the object block is referred to as a reference in the reference image of the object block MB. Fig. 39 shows the flow of processing of the motion vector calculation method of the present embodiment. First, it is judged whether or not the reference 15 block MB in the reference image following the target block MB has a motion vector (step S1). When the reference block MB does not have a motion vector (step S1; No), the motion vector is set to "0" to perform two-direction prediction (step S2) and the process of calculating the motion vector is ended. The reference block MB has a motion vector (Step SI; Yes), it is determined whether or not the reference block MB has a forward direction motion vector (step S3). 20 The reference block MB does not have a forward direction motion vector (step S3; No), since the reference block MB has only the rear direction The vector is moved, so the number of movement vectors in the subsequent direction is determined (step S14). In the case where the number of direction movement vectors after the reference block MB is "2", according to Fig. 19, Fig. 20, Fig. 21, and 22 In any of the calculation methods described in the figure, two direction predictions are performed using the two rear direction motion vectors which are scaled to 1,258,991, and the invention description (step S15). On the other hand, the number of vectors in the direction after the reference block MB is moved. In the case of "1", the unique back-direction direction motion vector of the reference block MB is scaled and the motion compensation is performed using the post-scaling direction motion vector (5 step S16). Once the step S15 or S16 is ended When the direction prediction is performed, the processing of the calculation method of the motion vector is ended. When the reference block MB has the forward motion vector (step S3; Yes), the number of motion vectors before the reference block MB is determined (step S4). Before the block MB is the number of vectors in the direction of movement of the "case Γ 10, it is determined with reference to the corresponding block before the party mobile mb whether the reference picture stored in the vector memory long-term or short-term memory (Step S5). In the case where the reference image corresponding to the previous motion vector of the reference block MB is stored in the short-term memory, the motion vector before the reference block MB is scaled and the direction is moved in the direction before the calibration is performed. Test (step S6). In the case where the reference image corresponding to the previous square motion vector of the reference block MB is stored in the long-term memory, according to the motion vector calculation method shown in FIG. 37, the motion vector before the reference block MB is not scaled and is originally localized. The two directions are predicted using the motion vector zero set in the backward direction (step 20 S7). When the prediction in the two directions of step S6 or step S7 is ended, the processing of the calculation method of the motion vector is ended. When the number of direction movement vectors after the reference block MB is "2", it is determined that the number of direction movement vectors before the reference image stored in the long-term memory in the forward direction movement vector of the reference block MB is determined (step S8). 76 1258991 玖 、 、 对应 对应 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 The belonging object image is scaled by the motion vector close to the chronological order and the two directions are predicted using the scaled movement to 毚5 (step S9). Corresponding to the reference image stored in the long-term memory, the direction moves to the number of turns in step S8, and the motion vector stored in the short-term memory is scaled and the scaled motion vector is used. The direction prediction is performed (step S10). 10 In the case where the direction is shifted to "2" in step S8 before the reference image stored in the long-term memory, the two directions are moved to the two sides to determine whether or not Referring to the same image in the long-term memory (step SU). In the case of referring to the same ceramic image in the long-term memory by moving the vector in the two front directions (step S11; Yes), the mobile vector stomach ten according to the first FIG. The Respect 15 method uses two motion vectors that are encoded or decoded in the image in the forward direction of the reference long-term memory to perform two directions (step S12). When both sides of the motion vector are referred to the same image in the long-term memory (step S11; No), according to the 20-movement vector calculation method described in FIG. 38, the corresponding storage in the long-term memory allocation is used. The two-direction prediction is performed by shifting the vector in the direction of the small image (step S13). The long-term memory stores the reference image regardless of the time of the actual image, so that the order of each reference image is allocated. The selection should be used to predict the direction of the moving direction in the two directions. Also, the order of the reference images stored in the long-term memory 77 1258991 玖, # may also coincide with the time of the image, but may be only with the memory. The order of the addresses is the same. That is, the order of the images stored in the long-term memory may not necessarily coincide with the time of the image. When the two directions of steps S12 and S13 are ended, the motion vector is ended. (Processing Method 11) Hereinafter, the embodiment U of the present invention will be described in detail with reference to the drawings. Fig. 40 is a block diagram showing the configuration of the moving image encoding device 1100 of the embodiment of the present invention. The moving picture coding apparatus may perform a spatial prediction method in which a direct mode is applied in the case where a coded block is constructed by a field and a coded block is constructed by a frame. A moving picture coding apparatus that performs motion picture coding includes a frame memory 11〇1, a difference calculation unit 1102, a prediction error coding unit 1103, a code sequence generation unit 1104, a prediction error decoding unit 1105, an addition unit 1106, and a frame memory. The body 1107, the motion vector detecting unit 1108, the mode selecting unit 1109, the encoding control unit H10, the switch 1111, the switch 1112, the switch 1113, the switch 1114, the switch 1115, and the motion vector memory m6. The frame memory 1101 is in image units. The image memory of the input image is held, and the difference calculation unit 1102 obtains and outputs a prediction error of the difference between the input image 20 from the frame memory 1101 and the reference image obtained from the decoded image based on the motion vector. . The prediction error coding unit 1103 quantizes and outputs the prediction error obtained by the difference calculation unit 11〇2, and outputs it. The code sequence generating unit 1104 performs variable length encoding on the encoding result from the prediction error encoding unit 1103, and converts it into a format of the output encoded bit stream. 78 1258991 、—, invention description—-------- ---------------------- ---------------------------- -------------, with the addition of information such as the front information, etc., which have been associated with the encoded prediction error, and the generation of the code sequence. The prediction error decoding unit 1105 performs variable length decoding on the coding result from the prediction error coding unit 1103, performs inverse quantization, and performs inverse frequency transformation such as IDCT conversion to decode the prediction error. *

The addition unit 1106 adds the reference image to the prediction error of the decoding result, and outputs a reference image indicating the input image and the image of the same frame as the encoded and decoded image data. The frame memory 1107 holds the image memory of the reference image in image units. The H motion vector detecting unit 1108 detects the motion vector for the coding unit of each coding target frame. The mode selection portion 1109 selects to calculate the movement direction amount in the direct mode or in other modes. The encoding control unit 1110 changes the image of the input image stored in the chronological order of the input frame memory 1101 in the encoding order. Further, the encoding control unit 1110 determines whether the unit of the predetermined size of each encoding target frame is encoded by the field structure or by the frame structure. Here, the unit of the predetermined size is set such that a macro block (for example, horizontal 16 pixels and vertical 16 pixels) is connected in the longitudinal direction (hereinafter referred to as a macro block pair). If the code is coded in the field structure, the frame memory 1101 reads the pixel 每隔 every other horizontal scanning line corresponding to the interlace, and if it is encoded by the frame structure, the input image is sequentially read from the frame memory 1101. Each of the image images of the image is placed on the memory by reading the pixel pairs of the coding target macro region corresponding to the field structure or the frame structure. The motion vector memory 1116 maintains a reference index of the motion vector of the macroblock and the frame to which the motion vector is to be referenced. The reference index is to keep the macroblocks in the encoded macroblock pair. 79 1258991 玖, invention description ————— --------------------------------------- ------------------ Next, the operation of the moving picture coding apparatus n〇〇 configured as described above will be described. • The input image is input to the frame memory 1101 in chronological order in image units. Fig. 41 (4) shows the order in which the moving image encoding device 1100 is input in chronological order in image units. Fig. 41(b) shows the order in which the arrangement of the 5th arrangement of the frame shown by the wth picture (&) is replaced by the coding order. In the vertical line of Fig. 0), there is no image, and the letter _ of the i-th word of the symbol shown at the lower right of each image indicates the image type (I, p or B), after the second word The numbers represent chronological image numbers. Further, Fig. 42 shows a configuration for explaining the reference frame list 300 of the embodiment U. The respective images of the input frame memory 11〇1 are alternately arranged in the encoding order in accordance with the encoding control unit 1110. The replacement order is performed in accordance with the reference relationship of inter-picture predictive coding, and the arrangement is replaced so that the picture used as the reference picture is encoded earlier than the picture used as the reference picture. For example, the 'P image is used as a reference image in one frame of the I or P 15 image two frames in the vicinity of the front in the display time order. In addition, the B image is displayed in chronological order in the vicinity of the front or one of the three frames of the p image and one frame in the vicinity of the rear or the p image in the chronological order. use. Specifically, the image P after the input image B5 and the image_B6 in FIG. 41 is referred to by the image B5 and the image B6, so that the image is replaced by the image B5 and the image. Like before B6. Similarly, the image P10 after the input image B8 and the image B9 is replaced by the image P8 and the image B9, and the image P13 after the input image B11 and the image B12 is replaced by the row. Before image B11 and image Bl2. Thereby, the result of replacing the image of Fig. 41 (a) is shown in Fig. 7 (7). 80 1258991 疚 - Each image that is alternately arranged by the frame memory 1101 is set to read out the unit of the macro block pair connecting the macro blocks in the vertical direction, and each macro block is set to the level 16 Pixel X is vertical 16 pixels in size. Therefore, the macro block pair constitutes a horizontal 16 pixel X vertical 32 pixel size. The encoding processing of the image 5 B11 will be described below, that is, the management of the reference frame list is performed by the encoding control unit 1110. Since the image B11 is a B image, inter-picture predictive coding using two directions is performed. The image B11 is used as a reference image in which two images of the images P10, P7, and P4 on the front side and the image P13 on the rear side in the display time order are displayed in the display time order. And it is set to specify any two of the four images among the four images in the macro block unit. Here, it is explained that the reference index is allocated in the initial state method. That is, the reference frame list 300 at the time of encoding of the image B11 is as shown in Fig. 42. In this case, the reference image is designated by the first reference index of Fig. 42 and the second reference image is designated by the second reference index of Fig. 42. In the processing of the image B11, the encoding control unit 1110 controls the switches so that the switch 1113 is turned "ON", and the switch 1114 and the switch 1115 are turned "OFF". Thus, the macroblock pair of the image B11 read from the frame memory 1101 is input to the motion vector detecting unit 1108, the mode 20 selecting unit 1109, and the difference calculating unit 1102. The motion vector detecting unit 1108 uses the decoded image data of the image 10, the image P7, the image P4, and the image P13 stored in the frame memory 1107 as a reference image, and detects the inclusion in the macro region. The first motion vector and the second motion vector of each macroblock of the block pair. The mode selection unit 1109 determines the coding mode of the macroblock pair using the motion vector detected by the motion vector detecting unit 1258991 and the invention description 1108. Here, the coding mode of the B picture can be set, for example, to be selected from intra-picture coding, inter-picture predictive coding using one direction motion vector, inter-picture predictive coding using two-direction motion vectors, and direct mode selection. . Further, in the case of selecting an encoding mode other than the direct mode, it is determined whether or not the macro block pair is encoded by the frame structure or encoded by the field structure. Here, a method of calculating a motion vector using a spatial prediction method of a direct mode will be described. Fig. 43(a) is a diagram showing an example of a motion vector calculation sequence using a prediction method in a 10 direct mode space in which a macroblock block paired with a field structure code and a macroblock block coded by a frame structure are mixed. flow chart. Fig. 43(b) is a diagram showing an example of the configuration of the peripheral macroblock pairing to which the present invention is applied in the case where the encoding target macroblock pairing is frame-coded. Fig. 43(c) shows an example of the configuration of the peripheral macroblock pairing applied in the present invention in the case where the coding target macroblock pairing is encoded by the field structure. The macro block pairs indicated by oblique lines in Fig. 43 (b) and Fig. 43 (c) are coded object macro block pairs. In the case where the coding object macroblock is coded using the spatial prediction of the direct mode, the three encoded macroblock pairs around the coding object macroblock pair may be selected. In this case, the encoding object macroblock pair can be encoded in any of the 20 field constructs or the box construct. Therefore, the encoding control unit 1110 first decides whether or not to perform the encoding target macroblock pair in the field structure or in the frame structure. For example, in the case where there are many field structure coders among the neighboring macroblock pairs, the coding object macroblock pair is encoded by the field structure, and in the case where there are many coders in the frame structure, the frame structure is performed. coding. For example, 82!25899i, by using the information of the surrounding blocks, it is decided whether to encode the encoding object macroblock pair in a frame structure, or to encode the field structure, and it is unnecessary to represent The information for encoding the coding target macroblock pair is described in the code sequence, and the appropriate structure can be selected from the bound macroblock pair prediction structure 5'. Next, the motion vector detecting unit 1108 calculates the motion vector of the encoding target macroblock pair in accordance with the determination by the encoding control unit 1110. First, the motion vector detecting unit 1108 detects whether or not encoding has been determined by the field structure (S301), or has determined that the encoding is performed in the frame structure, and if it is determined to be encoded in the frame structure, the encoding object macro is detected in the frame structure. The block pair motion vector (S302) detects the motion vector of the coding target macroblock pair in the field structure if it is determined to be in the idle state of the field structure coding (S303). Fig. 44 is a diagram showing the data composition of the macroblock pairing in the case of the frame structure coding and the data composition of the macroblock in the case of the field structure coding. 15 In the figure, white circles represent pixels on odd horizontal scan lines, and black circles with oblique lines indicate pixels on even horizontal scan lines. In the case where a macroblock pair is cut out from each frame representing the input image, pixels on the odd horizontal scanning line and pixels on the even horizontal scanning line are alternately arranged in the vertical direction as shown in the center of Fig. 44. In the case where such a macroblock pair is coded 20 in a frame structure, the macroblock is processed for every two macroblocks MB1 and macroblocks MB2, and respectively for the pair of macroblocks The two macroblocks MB1 and the macroblock MB2 obtain a motion vector. Further, in the case of the field structure coding, the macro block pair is divided into a macro block TF indicating a top field when the horizontal scanning line direction is staggered, and a macro block BF, 83 1258991 indicating the bottom field, The invention discloses that the motion vector system obtains one of the two field fields constituting the pair of macroblocks, respectively. On the premise of such a macroblock pair, as shown in Fig. 43(b), a case where the coding target macroblock is encoded in a frame structure will be described. Fig. 45 shows a more detailed processing sequence of step S302 shown in Fig. 43. Further, in the figure, a macroblock pair is indicated by MBP, and a macroblock is represented by MB. The mode selecting unit 1109 first calculates a motion vector by using the prediction of the spatiality of the direct mode for one macroblock MB1 (the upper macroblock) constituting the encoding target macroblock pair. First, the mode selection unit 1109 10 obtains the minimum value among the indices of the images of the neighboring macroblock pairs for the indices of the first motion vector and the second motion vector (S501). However, in this case, when the surrounding macroblocks are coded in a frame structure, only the macroblocks of the macroblocks of the adjacent coding objects are used. Secondly, it is detected whether the surrounding macroblock is encoded by the field structure (SS02), and in the case of the field structure coding, 15 is further detected from the reference frame list of FIG. 42 to form the peripheral macroblock. Among the fields in which the macroblocks are referenced, how many fields are the fields of the additional minimum index (S503). When the result of the detection in step S503 is that the field referenced by the two macroblocks is the field with the smallest index (ie, the same index), the average of the two motion vectors of the two macroblocks is obtained and is set to The surrounding macroblock moves the vector (S5〇4). This is due to the fact that in the case of the interlaced structure, the macroblocks in the frame structure are adjacent to the two macroblocks of the macroblocks adjacent to the field structure. The result detected in step S503 is only borrowed from a macroblock and referenced to 84 1258991 发, hair---------------------------- -------------------------------------------------- ---------------------- In the case where the field is attached with the smallest index, the motion vector of the one macro block is set as the peripheral macro block. Move vector (S504A). If the field to be referred to is a field in which the minimum index is not attached, the moving vector of the peripheral macro block is set to "〇" (S505). 5 In the motion vector of the surrounding macroblock, the motion vector with higher coding efficiency can be selected by using only the motion vector of the field with the smallest index added. The processing of S505 represents a case where it is judged that there is no motion vector applied to the prediction. The result detected in step S502 is a case where the peripheral macroblock pair is encoded by the frame structure, and among the peripheral macroblock pairs, the motion vector of the macroblock adjacent to the coding target macroblock is adjacent. The motion vector of the neighboring macroblock pair is set (S506). The mode selection unit 1109 repeats the processing from the above-described step S501 to step S506 with respect to the selected two neighboring macroblock pairs. The result is obtained for a macroblock in the pair of object macroblocks, for example, the motion vectors of the three surrounding macroblocks are respectively obtained for the macroblock MB1. Next, the mode selection unit 1109 detects whether the frame having the smallest reference index among the three surrounding macroblocks or the field in the frame is one (S507). In this case, the mode selection unit 1109 sets three peripheral macroblocks. The reference index is unified into a reference frame index or a reference field index and compared. In the reference frame list shown in FIG. 42, only the reference index is attached to each frame. However, the reference frame index has a certain relationship with the reference frame index in which only the index is attached in each frame, so that the reference frame can be borrowed from the reference frame. The list or the reference field list is calculated to be converted into the other reference index. 85 1258991 玖,_尔萌 Figure 46 shows the relationship between the reference field index and the reference frame index. As shown in Fig. 46, in the reference field list, the number of frames indicated by the first field and the second field f2 is along the time series, and each frame includes a frame of the coded 5 object block (in Fig. 46). The frame shown in the figure) is assigned a reference frame index of 0, 1, 2, . In addition, the first field fl and the second field f2 of each frame are allocated to 第, 1, 2 based on the first field fl including the frame of the encoding target block (the first field is the encoding target field). · · · Reference field index. Further, the reference field index is from the 10th field fl and the second field f2 of the frame close to the coding target field 1, and if the coding target block is the first field fl, the first field fl is preferentially allocated, and if it is the coding target area When the block is the first field fl, the first field is preferentially assigned. For example, when the surrounding macroblock of the frame structure code refers to the frame of the reference frame index "1", and the surrounding macroblock of the field structure code refers to the first field fl of the index "2" of the reference frame 15, the above The surrounding macroblocks are all processed with reference to the same image. That is, the reference frame index of the frame referenced by one peripheral macroblock satisfies one-half of the reference field index of the reference field to be allocated as another peripheral macroblock (rounded down to the decimal point) When the prerequisites are met, the surrounding macroblocks are processed with reference to the same image. For example, the coding target block of the first field fl indicated by Δ in Fig. 46 refers to the first field fl of the reference field index "2", and the surrounding macroblock of the frame structure refers to the reference frame index. In the case of the 1" frame, in order to satisfy the above conditions, the peripheral macroblocks are processed by referring to the same image. On the other hand, when a certain surrounding macroblock refers to the first field 86 1258991 fl of the reference frame index "2", and another surrounding macroblock refers to the frame of the reference frame index "3", in order to satisfy the above conditions, The peripheral macroblock is processed with reference to the same image. As described above, the result of the detection in step S5〇7, if one, 5 will refer to the frame with the smallest index or the surrounding macro of the field in the frame. The motion vector of the block is set as the motion vector of the coding target block (S508). If the result of the detection in step S507 is not one, it is further detected whether the frame with the smallest index among the three neighboring macroblock pairs or the macroblock block of the field in the frame is two or more (S509) If there are two or more, if one of the frames has the smallest reference frame or the surrounding macroblock of the field in the frame, the motion vector is set to "0" or more (S510). The center of the three motion vectors of the neighboring macroblocks is set as the motion vector of the encoding target block (S511). Further, if the result of the detection in step S509 is less than two, the movement vector of the coding target block is set when the reference index is the smallest frame or the number 15 of the peripheral macroblock pairs of the field in the frame is "0". Set to "0" (S512). As a result of the above processing, a macroblock of the encoding object macroblock pair is formed, for example, a motion vector MV is obtained for MB1 as a calculation result. The mode selection unit 1109 also performs the above-described processing on the motion vector having the second reference index and uses two obtained motion vectors to predict motion compensation in two directions. However, in the case where there is no peripheral macroblock of the first or second motion vector among the peripheral macroblocks, the motion vector of the direction is not used, and only the motion vector of one direction is used for motion compensation. Also, for another macroblock within the encoding object macroblock pair, for example 87 1258991 玖--------------------------- ---------------------------------------------- For the giant area Block MB2 also repeats the same processing. As a result, motion compensation is performed in a direct mode for two macroblocks of a coded object macroblock pair. Next, as shown in Fig. 43(c), it is explained that the coding object has a large set of block pairs encoded by the field structure. Fig. 47 is a flow chart showing the processing sequence of the more detailed step S303 shown in Fig. 43. The mode selection unit 1109 uses the spatial prediction of the direct mode for one macroblock constituting the coding target macroblock pair, for example, for the macroblock TF corresponding to the top field of the coding target macroblock pair. Calculate a motion vector MVt. First, the mode selection 10 unit 1109 finds the smallest among the indexes of the image to be referred to by the encoding target macroblock (S601). However, in the case where the field structure is used to process the neighboring macroblock pair, only the macroblock of the same field (top field or bottom field) as the encoding object macroblock is considered. Secondly, it is detected whether the encoding object macroblock pair is encoded in a frame structure (S6〇2), and if it is encoded in a frame structure, it is further determined according to the index of each frame given by the 15 reference frame list 300. Whether the frame referenced by the two macroblocks in the pair of surrounding macroblocks is attached to the frame of the minimum index (S603). As a result of the detection in step S603, when the frames referenced by the two macroblocks are the smallest index, the average of the motion vectors of the two macroblocks is determined to be 20値, and the coding object macroblock is set. The moving vector of the pair (S604). If the result of the detection in step S603 is that one or both of the frames that have been referenced do not have a frame with a minimum index, it is further detected whether the frame referenced by any macroblock does not have a minimum index (S605), and the detection is performed. The result is that in the case where the frame of reference 1 of any one of the macroblocks is attached to the minimum index, 88 1258991 κι^Wmm 'the motion vector of the macroblock is used as the motion vector of the pair of peripheral macroblocks. (S606), in the case of the detection in step S605, any macroblock is also set to the minimum index in the frame to be referenced, and the peripheral macroblock is set to "0" (S607). ). Among the moving motion vectors of the above-mentioned peripheral macroblocks, the frame to be referenced only uses the motion and vector of the frame attached to the minimum index, whereby a motion vector with higher coding efficiency can be selected. The processing of S607 represents the case where it is judged that there is no motion vector applied to the prediction. Further, in the case of the result of the detection in step S602, if the peripheral macroblock pair is encoded by the φ field structure, then the peripheral macroblock is The motion vector of the surrounding giant 10-set block pair is used as the motion vector of the macroblock corresponding to the target macroblock in the encoding target macroblock pair (S608). The mode selection 邰1109 is selected. The three peripheral macroblock pairs repeat the processing from step S601 to step S608. As a result, the motion vectors of the three surrounding macroblocks are respectively 15 blocks of a macroblock in the encoding object macroblock pair. For example, one for each of the macroblocks TF is obtained. Next, the motion vector detecting unit ιι 8 detects whether the macroblock pair of the frame with the smallest index among the three neighboring macroblock pairs is one. (S609), if it is one, the motion vector of the neighboring g set block pair of the frame whose reference index is the smallest is set to the moving direction of the encoding target macroblock pair (S610). The detection is performed in step S609 4 The result is not one Further, among the three neighboring macroblock pairs, whether the neighboring macroblock pair with the smallest index is two or more (5611), and if there are two or more, the unreferenced index is further The moving vector of the neighboring macroblock pair of the smallest frame is set to "〇" (5612), and the center of the three moving vectors of the surrounding macroblock is made to 89 1258991. The invention is described as the encoding object macro. The movement vector of the block (S613). Further, if the result of the detection in step S611 is less than two, since the number of the surrounding macroblocks of the frame whose reference index is the smallest is "0", the motion vector of the coding target macroblock is set to " 〇" (S614). 5 As a result of the above processing, a macroblock constituting a pair of coding target macroblocks obtains a moving vector MVt as a calculation result, for example, for the macroblock TF corresponding to the top field. The mode selection unit 1109 repeats the above processing for the second movement vector (second reference index). In this way, the motion vector formed by the two directions is used to obtain two motion vectors for the macroblock TF. However, in the case where there is no peripheral macroblock having the first or second motion vector among the peripheral macroblock pairs, the motion vector in the direction is not used, and only the motion vector in one direction is used for motion compensation. This is because the situation of the surrounding macroblocks referring to only one direction is only referring to one direction for the coding target macroblock, and the knowledge can be improved. Further, another macroblock in the pair of the coding target macroblock, for example, repeats the same processing for the macroblock BF corresponding to the bottom field. The result is a process consisting of two macroblocks in a pair of coding object macroblocks, for example, a macroblock TF and a macroblock BF. 20, in the case where the coding structure of the coding target macroblock pair is different from the coding structure of the surrounding macroblock pair, the average of the motion vectors of the two macroblocks in the pair of surrounding macroblocks is obtained. However, the present invention is not limited thereto. For example, the encoding object macroblock pair and the surrounding macroblock pair may be used only in the case of the same encoding structure, and the surrounding giant 90 908991 玖 may be used. The motion vector of the block pair is used, and in the case where the coding object macroblock pair is different from the neighboring macroblock pair structure, the motion vector of the neighboring macroblock pair having different coding structures may not be used. More specifically, first, in the case where 1 encoding object macroblock pair is encoded in a frame structure, only the motion vector of the peripheral macroblock pair of the frame structure code 5 is used. At this time, in the case where the motion vector of the peripheral macroblock pair coded by the frame structure has no reference index as the smallest motion vector, the motion vector of the coding target macroblock is set to "0". Further, when the surrounding macroblock encodes the field structure, the motion vector of the surrounding macroblock is set to "0". Next, in the case where the coding object macroblock is encoded in the field structure, only the motion vector of the peripheral macroblock pair encoded by the field structure is used. At this time, in the case where the motion vector of the surrounding macroblock of the field structure coding has no moving index with the smallest reference index, the motion vector of the coding target macroblock is set to "0". Further, in the case where the surrounding macroblock is coded in the frame structure, the moving vector of the surrounding macroblock 15 is set to "〇". After performing the calculation of each of the peripheral macroblock pairs in this way, 3, in the case of the motion vectors, referring to the frame having the smallest index or the field thereof, the motion vector is set to the direct mode. In the case of the motion vector of the coding object macroblock pair, if not, the center of the three motion vectors is set to the motion vector of the coding object macroblock pair in the direct mode. Moreover, the above description determines whether to encode the coding target macroblock pair by the field structure based on the majority of the encoded surrounding macroblock coding structure, or to block the coding object macroblock by the frame structure. It is coded, but the invention is not limited to this. For example, in the direct mode, the frame must be 91 1258991, the code must be encoded by the squirrel, or the field structure must be coded to determine the first fixedness. . In this case, for example, in the case where each encoding target is encoded in a frame structure or encoded in a field structure, it is described in the header of the entire code column or the front of each frame. The unit of switching may be, for example, a sequence, a 5 G 〇 P, an image, a segment, etc., and in this case, it may be described before the corresponding head in the code column. In this case, of course, only in the case where the coding object macro-region, the ghost pair, and the question-edge macroblock have the same coding structure, the direct mode can be calculated by using the method of moving the vector of the surrounding macroblock. A motion vector that encodes pairs of masculine macroblocks. Further, when transmitting in a package or the like, the front head portion and the data portion can be separated and individually transmitted. In this case, the front head and the data unit form a bit stream. However, in the case of a packet, the order of transmission is only before and after the other part of the data part corresponding to the corresponding data part is transmitted, even if it does not constitute a bit stream. In this way, by the case where the frame structure is used or the field structure 15 is used in a fixed manner, the process of determining the structure using the surrounding block information can be reduced, and the simplification of the process can be achieved. Further, in the direct mode, it is also possible to use both the frame structure and the field structure to process the coding target macroblock pair and use the 'method of selecting a structure with high coding efficiency. In this way, only in the case where the coding object macroblock pair and the surrounding macroblock are the same coding structure, the direct mode can be calculated by using the method of moving the vector of the surrounding macroblock. The vector of the encoded object macro ^ block pair. By using this method, although it is indicated that the information of one of the frame structure and the field structure is selected in the code column, the residual signal of the motion compensation can be further reduced to improve the coding efficiency. 92 1258991 玖, 翻面 Further, in the above description, the situation in which the peripheral macroblock is moved and compensated by the size of the macroblock is described. However, the motion compensation can be performed in different sizes. In this case, as shown in Fig. 48 (a) and (b), for each macroblock of the encoding target macroblock pair, the movement of the block including the pixels at the positions a, b, and 5 c will be moved. The vector is set to the motion vector of the neighboring macroblock pair. Here, Fig. 48(a) shows a case where the lower macroblock is processed. In the case where the coding object macroblock is different from the frame/field structure of the surrounding macroblock, the area including the pixels at the positions a, b, and c shown in (a) and (b) of Fig. 49 is used. The block is processed in 10 rows with the blocks of the positions a', b', c'. The positions a', b', and c' are blocks of the other macroblock included in the same macroblock corresponding to the positions of the pixel positions a, b, and c. For example, in Fig. 49(a), in the case where the coding target macroblock pair is different from the frame/field structure of the surrounding macroblock pair, the motion vector of the block on the left side of the upper coding target macroblock is BL1. Determined by the motion vector of BL2. 15 Further, for example, in FIG. 49(b), in the case where the coding target macroblock pair is different from the frame/field structure of the surrounding macroblock pair, the motion vector of the block on the left side of the upper coding target macroblock is It is determined by using the motion vectors of BL3 and BL4. With such a processing method, it is possible to perform a direct mode processing in which the difference between the 20 frames and the field is considered in the case where the peripheral macroblock is subjected to motion compensation in units different from the size of the macroblock. Further, in the case where the peripheral macroblock performs motion compensation in units of a size different from the size of the macroblock, the average value of the motion vector of the block included in the macroblock can also be obtained as the The motion vector of the macro block. In the case where the peripheral macroblock is used for motion compensation in units different from the size of the macroblock, the direct mode processing of the difference between the frame and the field can be performed. As described above, the result of detecting the motion vector and performing inter-picture predictive coding according to the detected motion vector is based on the motion vector detected by the motion vector detecting unit 1108 and the encoded prediction error image in each macro region. Block 5 is stored in the code train. However, for the motion vector of the macroblock coded in the direct mode, only the case of encoding in the direct mode is described, and the motion vector and the reference index are not described in the code sequence. The 50th graph is not an example of the data structure of the code sequence 7〇0 generated by the code sequence generating unit 11〇4. As shown in the figure, in the code sequence 700 generated by the code generating portion 11〇4, a head leader is provided for each image Picture. In the previous Header, an item RPSL indicating a change of the reference frame list 10 and an item not shown in the figure indicating the image type of the image are provided, and the item RPSL is changed from the initial setting to the first reference of the reference frame list 10. In the case of the allocation method between the index 12 and the second reference index 13, the changed allocation method will be described. On the other hand, the encoded prediction error is recorded in each macroblock. For example, in the case where a certain macroblock is coded using the spatial prediction of the direct mode, the item Block1 describing the prediction error corresponding to the macroblock does not describe the motion vector of the macroblock, and is described in the representation. The PredType encoding mode of the encoding mode of the macroblock is the information of the direct mode. Further, in the case where it is selected from the viewpoint of coding efficiency as described above whether or not the macroblock is encoded in one of a frame structure or a field structure, information indicating whether one of a frame structure or a field structure is selected is described. The 'coded prediction error' is then described in the project CodedRes. In addition, when the other macroblocks are macroblocks coded by the inter-picture predictive coding mode, the code is described in the item B1〇ck2 corresponding to the prediction error of the 1258991 [the budding macroblock] The pattern of the pattern PredType describes that the coding mode of the macroblock is described as an inter-picture prediction coding mode. In this case, in addition to the coding mode, the first reference index 12 of the macroblock is written to the item Ridx0, and the second reference index 13 is written to the item 5 Ridxl. The reference index in the block appears as a variable length codeword and is smaller and smaller the code length of the code. Further, the moving vector description item MV0 is displayed next to the block before the macro block, and the moving vector is described in the rear block reference item MV1. Then, the encoded prediction error is described as an item.

CodedRes 〇 10 Fig. 51 is a block diagram showing the configuration of the dynamic image decoding apparatus 800 for decoding the code sequence 700 shown in Fig. 50. The moving picture decoding device 8 includes a moving picture decoding device that decodes the code sequence 700 including the prediction error of the macroblock coded in the direct mode, and has a code sequence analysis unit 701 and a prediction error decoding unit 702. The mode decoding unit 703, the motion compensation decoding unit 〇5, the motion vector storage unit 〇6, the frame memory 〇7, the addition unit 708, the switch 709 and the switch 710, and the motion vector decoding unit 711. The code string analysis unit 701 extracts various materials from the input code sequence 700. The various data described herein are information on the coding mode and information about the motion vector. The information of the extracted coding mode is output to the mode decoding unit 703. Further, the information of the motion vector extracted by 2〇 is output to the motion vector decoding unit 711. Further, the extracted prediction error coded data is output to the prediction error decoding unit 702. The prediction error decoding unit 702 performs decoding of the input prediction error coding material to generate a prediction error image. The resulting predicted error map is output to switch 709. For example, when the switch 7〇9 is connected to the terminal b, the prediction error 1258991 玖 and the fading image are output to the adder 708. The mode decoding unit 703 performs control of the switch 7〇9 and the switch 710 with reference to the coding mode information extracted from the code sequence. In the case where the encoding mode is intra-picture encoding, it is controlled to connect the switch 709 to the terminal a and connect the switch 701 to the terminal c. In the case where the encoding mode is inter-picture encoding, it is controlled to connect the switch 709 to the terminal b and the switch 701 to the terminal d. Further, the mode decoding unit 703 also outputs the information of the coding mode to the motion compensation decoding unit 705 and the motion vector decoding unit 711. The motion vector decoding unit 711 performs decoding processing on the encoded motion vector input from the code 10 column decoding unit 610. The decoded reference picture number and the motion vector are held in the motion vector memory unit 706 and output to the motion compensation decoding unit 705. When the encoding mode is the direct mode, the mode decoding unit 703 controls the switch 709 to be connected to the terminal b to connect the switch 701 to the terminal d. Further, the mode decoding unit 〇3 also outputs the information of the coding mode to the motion compensation decoding unit 705 and the motion vector decoding unit 711. When the encoding mode is the direct mode, the motion vector decoding unit 711 determines the motion vector used in the direct mode using the motion vector and the reference image number stored in the peripheral macroblock pair of the mobile vector memory unit 706. The determination of the motion vector 20 is the same as that described in the operation of the mode selection unit 1〇9 of Fig. 40, and therefore the description thereof will be omitted. The motion compensation decoding unit 705 acquires motion compensation from the frame memory 707 for each macroblock based on the decoded reference picture number and the motion vector. The obtained motion compensation image is output to the addition unit 708. Frame Memory 96 1258991 玖—, Flip — The 707 system holds the decoded image in the memory of each frame. The addition unit 708 adds the input prediction error image and the motion compensation image to generate a decoded image. The generated decoded image is output to the frame memory 707. As described above, according to the embodiment, in the spatial mode prediction method of the direct mode, the coding macroblock block is encoded in the surrounding macroblock, and the motion vector coded in the frame structure is mixed. The motion vector can also be easily obtained in the case of a motion vector encoded in the field structure.

Further, in the above-described embodiment, each image is processed by using any one of a frame structure or a 10-field structure in which two macroblocks are connected to a unit of a macroblock pair in the vertical direction. In the case, however, the switching frame configuration or the field configuration can also be performed in different units, for example, in macroblock units.

Further, in the above-described embodiment, the case where the macroblock in the B image is processed in the direct mode has been described, but the same 15 processing can be performed on the P image. In the encoding and decoding of the P picture, each block is only motion compensated from one picture, and only one reference frame list is used. Therefore, in the case where the same processing is performed on the p-picture, the processing of the two motion vectors (the first reference frame list and the second reference frame list) of the encoding/decoding block is obtained in the present embodiment. It is required to process a moving vector. 20 In the above embodiment, a case is described in which the motion vector used in the direct mode is predicted by using the motion vectors of the three neighboring macroblock pairs, but the number of neighboring macroblock pairs used may also be used. For different embarrassment. For example, consider the case of using only the motion vector of the neighboring macroblock pair of the left neighbor. 97 1258991 发明, invention description (implementation mode I2) A memory medium such as a flexible disk is recorded by a program for realizing the image coding method and the image decoding method shown in each of the above embodiments. The processing shown in each of the above embodiments can be simply implemented in an independent computer system 5. Fig. 52 is an explanatory diagram of a memory medium for storing a program for realizing an image encoding method and an image decoding method from the above-described embodiment to the embodiment 11 by a computer system. Fig. 52(b) shows the appearance, the broken 10-sided structure, and the flexible disk as seen from the front side of the flexible disk, and Fig. 52(a) shows the physical state of the flexible disk of the recording medium body. An example of the format. The flexible disk FD is housed in the cassette F, and the surface of the disk forms a plurality of tracks which are concentrically oriented from the outer circumference toward the inner circumference, and the magnetic tracks are divided into 16 sectors Se in each angular direction. Therefore, the field in which the flexible disk storing the above-described program is distributed on the flexible disk FD 15 is recorded as the image encoding method and the image decoding method of the above program. Further, Fig. 52(c) shows a configuration in which the flexible disk FD performs recording and reproduction of the above-described program. In the case where the above-described program is recorded on the flexible disk FD, the image encoding method and the image decoding method described above are written from the computer system Cs by the flexible disk drive. Moreover, when the image encoding method and the image decoding method are constructed in a computer system by a program in a flexible disc, the flexible disc drive is used to transfer the program from the flexible disc. Read and transfer to the computer system. Moreover, the above description explains the use of a flexible disc as a recording medium, 1258991 ........................... .....................-............................ .............................................. ....... 发明, invention instructions, however, the use of optical discs can be carried out in the same way. Further, the recording medium is not limited thereto, and the components of the recordable program such as a CD-ROM, a memory card, and a ROM cassette can be similarly implemented. Further, an application example of the image coding method and the image 5 decoding method shown in the above embodiment and a system using the same will be described. Fig. 53 is a block diagram showing the overall configuration of the content supply system exl〇〇 which realizes the content distribution service. The base station exl07 to exl10 of the fixed wireless station are separately provided in each unit cell by dividing the supply area of the communication service into a desired size. 10 The content supply system exioo is connected to a computer exlll, a PDA (personal digital assistant) exll2, a camera exll3, for example, via the Internet service provider 102 and the telephone network exl04, and the base stations exl〇7 to exllO. Mobile phone exll4, mobile phone with camera, exll5 and other machines. However, the content supply system ex100 is not limited to the combination of Fig. 53, and any of them may be combined and connected. Further, each device may directly connect to the telephone network ex104 without the base station ex107 to exllO of the fixed wireless station. Camera exll3 coefficient camera, etc. Machine that can change dynamic images. Further, the mobile phone is a PDC (Personal Digital Communications) 2 MIMO system, a CDMA (Code Division Multiple Access) system, a W-CDMA (Wideband-Code Division Multiple Access) system, or a GSM (Global System for Mobile Communications) system. A mobile phone, or a PHS (Personal Handyphone System) can be used in any way. 99 1258991 发明Inventive Description Further, the 'streaming server 103' is connected from the camera exl 13 through the base station exl〇9 and the telephone network exi〇4, and can be performed by using the camera exll3 according to the encoded data transmitted by the user. Delivery message. The encoding processing of the captured data can be performed by the camera exll3, or by a server that performs the processing of the data. Further, the moving image data photographed by the camera exll6 can also be transmitted to the streaming server 103 by the computer exlll. A camera that can capture still images and moving images, such as a camera exll6 coefficient camera. In this case, the encoding of the moving image data can be performed by any of the computer exlls in the camera exll6. Further, the 10-code processing is performed by LSI exll7 having a computer exlll or a camera exll6. Further, the image encoding/decoding software may be incorporated in any storage medium (CD-ROM, flexible disk, hard disk, etc.) of a readable recording medium such as a computer exiil. It is also possible to transmit dynamic image data on the mobile phone exll5 with a camera. The moving image data at this time is the data processed by the L SI encoding of the mobile phone eX115. The content supply system ex100 displays the content captured by the user with the camera exll3 and the camera exll6 (for example, a video of live music, etc.) in the same manner as in the above-described embodiment, and transmits the content to the streaming server ex 1〇3 at the same time. The traffic server exl03 allocates the above two contents to the requester. The requester has a computer exlll, a PDAexll2, a camera exll3, a mobile phone exii4, etc., which can decode the above-mentioned encoded data. Under such conditions, the content supply system exl 1 can receive and reproduce the flat code 2; and the system for personal playback can be realized by the reproduction. 100 1258991 1 . 1 , , s,.^ . r , ,, 发明, invention Description The encoding and decoding of each machine constituting the system is performed by using the image encoding device or image decoding shown in each of the above embodiments. The device is OK. Take a mobile phone as an example. Fig. 54 shows a mobile phone exll5 using the image coding method 5 and the image decoding method explained in the above embodiment. The mobile phone 115 has an antenna ex2〇1 for transmitting a reception wave with the base station exllO, a camera unit eX203 capable of capturing a video or a still picture such as a CCD camera, and an image captured by the camera unit ex203 and received by the antenna ex201. a display unit ex202 such as a liquid crystal display such as a video or the like, and a sound output unit ex208 such as a main unit configured to operate a 10-key ex204 group, a speaker for outputting sound, or the like, a sounder for inputting sound, or the like. a sound input unit ex205; a recording medium ex207 for storing the captured moving image or still picture material, the received mail material, the moving image data or the still image data, the encoded data or the decoded data; The recording unit I5 media ex2〇7 can be installed in the jack portion ex206 of the mobile phone exll5. The recording medium ex207 is a type of flash memory recording medium in which an EEPROM (Electrically Erasable and Programmable Read Only Memory) that can be electrically rewritten or erased is stored in a plastic case such as an SD card. 20 Further, the mobile phone exll5 will be described with reference to FIG. The mobile phone exll5 is connected to the power supply circuit unit ex310 and the operation input via the synchronous bus bar ex313 by the main control unit ex311 configured to integrally control the respective units including the display unit ex202 and the operation key ex204. Control unit ex304, image coding unit ex312, camera interface ex303 101 1258991, invention description, LCD (Liquid Crystal Display) control unit ex302, image decoding unit ex3〇9, demultiplexing unit ex308, and recording/reproduction unit ex3〇 7. The modulation and demodulation circuit unit ex306 and the audio processing unit ex305. When the power supply circuit unit ex310 is in a state in which the end dialogue and the power source key are turned ON by the user's operation, the battery pack is supplied with power from the battery pack, and the camera-equipped digital mobile phone exll5 is activated. The mobile phone exll5 converts the voice signal collected by the voice input unit ex205 into the digital voice data by the voice processing unit ex305 under the control of the main control unit ex311 such as the CPU, the ROM, and the RAM, and demodulates it by the variable tone. The circuit unit ex306 performs spectral diffusion processing on the digital audio data, performs digital contrast conversion processing and frequency conversion processing on the transmission/reception circuit unit ex301, and transmits the data by the antenna ex201. Further, the mobile phone exll5 amplifies the data received by the antenna ex201 in the voice call mode, performs frequency conversion processing and digital contrast conversion processing, and performs spectral inverse diffusion processing on the variable-modulation demodulation circuit unit ex306 to perform sound processing. After the part ex305 is converted into a contrast sound, the contrast sound is output by the sound output unit ex208. Furthermore, in the case of transmitting an e-mail in the data communication mode, the text data of the e-mail input by the operation of the operation key ex204 of the main body unit is sent to the main control by the operation input control unit ex304. Department ex311. The main control unit ex311 performs spectral diffusion processing on the data in the variable-modulation circuit unit ex306, and the gray-scale receiving circuit unit ex301 performs digital contrast conversion processing and frequency conversion processing, and then transmits 102 1258991 by the antenna ex201. To the base station exllO. When the image data is transmitted in the material communication mode, the image data captured by the camera unit ex203 is supplied to the image encoding unit ex312 by the camera interface ex303. Further, when the image data is not transmitted, the image data captured by the camera unit ex203 can be directly displayed by the camera interface ex303 and the LCD control unit ex3〇2. The image encoding unit ex312 has the configuration of the image encoding device described in the present invention, and the image data supplied from the camera unit ex203 is given by the φ encoding method of the image encoding device described in the above embodiment. The encoded code is converted into a coded image data, and the coded image data is sent to the demultiplexing unit ex308. At this time, the mobile phone exll5 transmits the sound collected by the camera unit ex2〇3 to the sound input unit ex205 during the shooting by the sound processing unit ex305 as digital sound data to the demultiplexing unit ex308. The demultiplexing unit ex308 multiplexes the encoded image data supplied from the image encoding unit ex312 and the sound data supplied from the audio processing unit ex305 in a predetermined manner, and the multiplexed data obtained as a result is changed. The demodulation circuit unit ex306 performs spectrum diffusion processing, performs digital-contrast conversion processing and frequency conversion processing on the transmission/reception circuit unit 'ex301, and transmits it by the day 20 line ex201. In the case of receiving the dynamic image file linked to the home page in the data communication mode, the received data received from the base station ex110 is spectrally de-diffused by the variable-modulation demodulating circuit unit ex306 by the antenna ex201. Processing, and the multiplexed data obtained by the result is sent to the multi-channel 103 1258991 发明, the invention description is ex308. Further, the multiplexed data received by the antenna ex2〇1 is decoded, and the demultiplexing unit ex3 〇8 separates the bit stream of the image data into the bit stream of the sound data by separating the multiplexed data. By supplying the coded image data to the image decoding unit ex309 and supplying the audio data to the audio processing unit ex305° by the synchronization bus 5 ex313, the image decoding unit ex309 has the map described in the present invention. In the configuration of the decoding device, the bit stream of the image data is decoded by the decoding method corresponding to the encoding method shown in the above embodiment, thereby generating the reproduced dynamic image 10 image data, and the LCD control unit ex302 The reproduced moving image data is supplied to the display unit ex202, and thus, for example, the reproduced moving image data included in the reproduced moving image file linked to the top page can be displayed. At the same time, the sound processing unit ex305 converts the sound data into the contrast sound data, and then supplies the contrast sound data to the sound output unit ex208, whereby the sound included in the reproduced moving image file linked to the top page can be reproduced. data. Further, the present invention is not limited to the above-described system example, and a broadcast composed of a satellite or a ground wave has recently become a topic. As shown in FIG. 56, at least an image coding apparatus or an image of the above-described embodiment can be incorporated in a digital broadcasting system. Any one of the decoding devices. Specifically, the image stream of the broadcast station ex409 transmits the bit stream of the information to the communication satellite or the broadcast satellite ex410 by radio waves. The broadcast satellite ex410 that receives this bit stream transmits a radio wave for broadcasting and receives the radio wave with an antenna ex406 of a home having a satellite broadcast receiving device, and borrows a television (receiver) ex4〇1 or a video cassette (STB). The ex407 temple device decodes the bit stream and regenerates it. Moreover, it is also possible to read the bit stream of the storage medium ex4〇2 such as the CD or DVD recorded on the recording medium, and to install the image decoding apparatus shown in the above embodiment on the reproduction to be decoded. Device ex4〇3. The image signal reproduced in this case is displayed on the monitor ex404. In addition, an image decoding device can be installed in the video cassette ex4〇7 of the cable ex4〇5 connected to the cable for cable television or the antenna ex4〇6 of the satellite 5/ground wave broadcast, and is configured to be a screen of the television ex408. The situation of regeneration. In this case, it is also possible to assemble the image decoding device in the television without assembling the image decoding device in the video-on-demand box. Further, the vehicle ex412 having the antenna ex411 receives the signal from the satellite ex41() or the signal from the base station exl77, and can reproduce the display device such as the onboard satellite navigation device ex413 of the vehicle ex412. Dynamic image. Further, the image coding apparatus shown in the above embodiment can encode the image signal and record it on the recording medium. A specific example is a recording of an image signal on a DVD recorder of a DVD disc ex421 or a recorder ex420 recorded on a disc recorder of the hard disk 15. And can be recorded on the SD card ex422. When the recorder ex420 has the image decoding apparatus shown in the above embodiment, the image signal recorded on the DVD disc ex421 or the SD card ex422 can be reproduced and displayed on the screen ex408. In addition, the configuration of the in-vehicle satellite navigation ex413 is such that, for example, the configuration of the camera unit ex203, the camera interface ex303, and the image encoding unit ex312 is excluded in the configuration shown in FIG. 5520, and the computer can be known in the same manner. Exlll or TV (receiver) ex4〇l and so on. In addition, the terminal of the mobile phone exll4 or the like has a transmitting and receiving terminal of both the encoder and the decoder, and it is known that only the terminal is 105 1258991, and the transmitting terminal of the invention code is only There are three installation forms of the decoder's receiving terminal. As described above, the moving picture coding method or the moving picture decoding method shown in the above embodiment can be used in any of the above-described apparatuses and systems, so that the effects described in the above embodiment can be obtained. Further, the present invention is not limited to the above-described embodiment, and various types of changes can be made without departing from the scope of the invention. Industrial Applicability: The image coding apparatus of the present invention is applicable to an image 10 coding apparatus such as a personal computer having a communication function, a PDA, a digital broadcasting station, and a mobile phone. Further, the image decoding apparatus of the present invention can be used for an image decoding apparatus such as a personal computer having a communication function, a PDA, an STB that receives digital broadcasting, and a mobile phone. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram for explaining a reference relationship of an image of a conventional example. FIG. 2 is a schematic diagram for explaining an operation of a direct mode of a conventional example (FIG. 3) a) is an example of a motion vector prediction method in the case of a front image in the B image reference time using the spatial prediction method 20 of the conventional direct mode. Fig. 3(b) shows an example of a reference image list in which each encoding target image is created. Fig. 4 (A) and (B) are explanatory diagrams of an image number and a reference index. 106 1258991 BRIEF DESCRIPTION OF THE DRAWINGS Fig. 5 is a view showing a format of an image coded signal format constituted by a conventional image coding apparatus. Fig. 6 is a block diagram for explaining the coding operation of the embodiment 1 and the embodiment 2 of the present invention. Fig. 7 is a schematic diagram for explaining the operation in the case where the block referring to the motion vector in the direct mode has a reference to display two motion vectors ahead in the chronological order. Fig. 8 (a.) and (b) are pattern diagrams for comparing the image reference relationship between the display order and the encoding order. Fig. 9 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. Fig. 10 (a) and (b) are diagrams for comparing the image display relationship of the display order and the coding order. Figure 11 is a block diagram for explaining the decoding operation of the embodiment 5 and the embodiment 6 of the present invention. Fig. 12 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. Fig. I3 is a schematic diagram for explaining the operation of referring to the motion vector in the direct mode with reference to the case of displaying two motion vectors in the chronological order. The second diagram is used to illustrate the block in the direct mode with reference to the motion vector. With the reference to display the two motion vectors in the rear in chronological order.

Claims (1)

1258991 拾,罕 IfWWfW ι· A motion vector calculation method, which is a motion vector calculation method for deriving a motion vector of a block to constitute an image of a moving image from the calculated motion vector of the peripheral block, characterized in that The method includes: in the case that the block to be exported is encoded or decoded in a frame structure, for the peripheral block coded or decoded by the field structure, the block of the motion vector and the bottom field of the block in the top field is used. Among the motion vectors, a motion vector that satisfies a certain condition is derived, and a motion vector that is the block to be derived is derived; in the case where the block to be derived is encoded or decoded in a field structure, encoding or decoding is performed in a frame structure. The peripheral block is derived from a motion vector of two blocks adjacent to each other, and a motion vector satisfying a certain condition is derived to derive a motion vector of the block to be derived. 2. The method of calculating a motion vector according to claim 1, wherein the certain condition is that the block of the top field is moved in a case where the block to be derived is encoded or decoded in a frame structure. The image in which the vector and the motion vector of the block of the bottom field respectively refer to the image having the so-called identical minimum reference index among the referenceable images; and the block to be the derived object is encoded by the field structure In the case of decoding, the image to which the motion vectors of the two adjacent blocks are respectively referred to has a condition of an image of a so-called minimum reference index among the images of FIG. 20 which can be referred to. 3. The method for calculating a motion vector according to claim 2, wherein, in a case where both of the motion vectors of the two blocks satisfy the predetermined condition, the average of the motion vectors of the two blocks is used to derive The motion vector of the block to be exported. The method of calculating a motion vector according to claim 3, wherein, in the case where the predetermined condition is not satisfied, the motion vector of the peripheral block is set to "〇" The motion vector that is the block of the aforementioned export object is derived. 5. The method of calculating a motion vector according to claim 1, wherein the certain condition is that the block of the top field is moved in a case where the block to be derived is encoded or decoded in a frame structure. The image of the vector and the motion vector of the block of the bottom field respectively refers to the condition of the image of the so-called same reference index among the images that can be referred to; and the 10 blocks that become the object of the export are encoded by the field structure or In the case of decoding, the images respectively referred to by the motion vectors of the two adjacent blocks have a condition of an image of the same reference index among the images that can be referred to. 6. The method of calculating a motion vector according to claim 5, wherein, in the case where both of the motion vectors of the two blocks satisfy the predetermined condition, the average of the motion vectors of the two blocks is used. The motion vector that is the block of the aforementioned export object is derived. 7. The method of calculating a motion vector according to claim 6, wherein, when the predetermined condition is not satisfied, the motion vector of the peripheral block is set to "〇", and the block to be exported is derived. Move to 20 quantities. 8. The motion vector calculation method according to claim 1, wherein the block to be derived is encoded or decoded in a frame structure, or encoded or decoded in a field structure, and is fixed in advance. 9. The method for calculating a motion vector according to item 8 of the patent application scope, wherein 115 1258991 picks up, examines the patent range. The motion vector calculation method derives a moving image from the calculated peripheral block to form a dynamic image. The motion vector of the block of the image is characterized in that the motion vector calculation method includes: in the case where the block to be the derived object is decoded by the frame structure, the peripheral block decoded by the field structure is used in Among the motion vectors of the block of the top field and the motion vector of the block of the bottom field, the motion vector satisfying the condition is derived and the motion vector of the block to be derived is derived; In the case of decoding, for the peripheral block decoded by the frame structure, the motion vector satisfying the condition is used in the motion vector of the two blocks adjacent to each other, and the motion vector of the block to be derived is derived. 118