CN102227132A - Motion vector prediction coding and decoding method, apparatus thereof and coding and decoding system - Google Patents

Abstract

The invention discloses a motion vector prediction coding and decoding method, an apparatus thereof and a coding and decoding system, relating to the video coding and decoding field. The method comprises the following steps: through constructing a specific motion vector candidate set, carrying out simulation decoding at a coding terminal according to the specific motion vector candidate set; comparing an optimal motion vector predicted value selected from the candidate motion vector set and an optimal motion vector analog value in the simulation decoding; determining whether to add a description indicating a position sequence number of the optimal motion vector analog value in the candidate motion vector set in an output code flow; through detecting whether received code flow data contains the motion vector position sequence number or not, employing different decoding method with a decoding terminal. According to the invention, through space and time correlation, predicted value of the motion vector is more accurately selected, code rate of coding motion vector difference is reduced, motion vector data can be more finely compressed, and output code rate is reduced.

Description

A kind of motion-vector prediction decoding method, device and coding/decoding system

Technical field

The present invention relates to the coding and decoding video field,, relate in particular to a kind of motion-vector prediction decoding method, device and coding/decoding system.

Background technology

Along with Video Applications is constantly carried out, video encoding standard and algorithm have also obtained fast development, and for example up-to-date video encoding standard H264 compares traditional coding standard such as H263 picture compression efficiency to be had greatly and promote, but the new problem of also having brought some simultaneously.The image encoding algorithm is meticulousr in the new video encoding standard, and is also more and more littler to the division of view data, and for example H264 is exactly 4x4 to the brightness data minimal processing unit.The advantage of doing like this is in the inter prediction process, can the lifter motion accuracy for predicting.Weak point is the increase that becoming more meticulous of data processing unit brought motion vector, need expend lot of data and come the transmitting moving vector, cause the increase of code check, this deficiency is especially outstanding in low bit-rate video compression, and about data of 20% to 30% are used for the transmitting moving vector in low bit-rate video stream according to statistics.

Summary of the invention

The purpose of the embodiment of the invention is to propose a kind of motion-vector prediction coding method, is intended to solve prior art in the low bit-rate video compression, and becoming more meticulous of data processing unit brought the increase of motion vector, causes the increase problem of code check.

A kind of motion-vector prediction coding method of the embodiment of the invention is achieved in that described method comprises:

Obtain motion vector (mv) through current block behind the motion prediction;

Structure is used for the set of candidate motion vectors of motion-vector prediction coding;

From set of candidate motion vectors, choose optimal motion vector predictor (mv _Best);

Ask for the motion vector difference (mvd) of current block motion vector and described optimal motion vector predictor.

Preferably, step is asked for and is also comprised step after the motion vector difference of current block motion vector and described optimal motion vector predictor:

At coding side motion vector is carried out analog codec, to obtain the optimal motion vector analogue value (mv _Dec).

Preferably, step is asked for and is also comprised step after the motion vector difference of current block motion vector and described optimal motion vector predictor:

(mvd) is sent to decoding end through the entropy coding module coding with described motion vector difference.

Second purpose of the embodiment of the invention is to propose a kind of motion-vector prediction code device, and described device comprises:

The motion vector acquisition module is used to obtain the motion vector through current block behind the motion prediction;

Set of candidate motion vectors makes up module, is used to make up the set of candidate motion vectors that is used for the motion-vector prediction coding; Be specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors

[mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors;

The optimal motion vector predictor is chosen module, is used for choosing the optimal motion vector predictor from set of candidate motion vectors, and it comprises motion-vector prediction difference bit number computing module, motion vector difference selection module;

Motion-vector prediction difference bit number computing module is used for the bit number R (mv-mv of calculation code motion-vector prediction difference needs _P), wherein, mv _P∈ [mv _AMv _BMv _CMv _H264Mv _DMv _E], wherein mv is the current block motion vector, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select;

The motion vector difference bit number is selected module, is used for selecting to make encoding motion vector difference bit number R (mv-mv _P) minimum motion vector predictor mv _PBe optimal motion vector predictor mv _Best

Motion vector difference is asked for module, is used to ask for current block motion vector mv and described optimal motion vector predictor mv _BestMotion vector difference mvd, mvd=mv-mv _Best

Preferably, described device further comprises:

The motion vector difference sending module is used for described motion vector difference mvd is sent to decoding end through the entropy coding module coding.

Preferably, described device further comprises:

The analog codec module is asked for module with described motion vector difference and is linked to each other, and is used at coding side motion vector being carried out analog codec, to obtain the optimal motion vector analogue value.

The 3rd purpose of the embodiment of the invention is to propose a kind of motion-vector prediction coding/decoding method, and described method comprises:

Received code end transmitted stream data;

Detect in the bit stream data that receives and whether comprise the position number of motion vector in set of candidate motion vectors;

According to whether comprising the motion vector position number in the bit stream data, decoding in different ways.

, described " according to whether comprising the motion vector position number in the bit stream data, adopting different modes to decode " is specially:

When comprising described position number in the described bit stream data, according to described motion vector position number, from the concentrated optimal motion vector predictor mv that obtains of motion vector candidates _Best, further decoding is calculated goes out current block motion vector mv, wherein, and mvd=mv-mv _Best

When not comprising described position number in the described bit stream data, the motion vector difference mvd that receives decoded calculates the current block motion vector, wherein, and mvd=mv-mv _Best

The 4th purpose of the embodiment of the invention is to propose a kind of motion-vector prediction decoding device, and described device comprises:

Data reception module is used for received code end transmitted stream data, and described bit stream data comprises motion vector difference and/or motion vector position number;

Motion vector position number detection module, whether the bit stream data that is used for detecting reception comprises the motion vector position number, and described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors;

Decoder module is used for when bit stream data does not comprise the position number of motion vector, and the motion vector difference that receives is decoded calculates the current block motion vector; Also be used for when bit stream data comprises the motion vector position number, according to described motion vector position number, concentrate from motion vector candidates and to obtain the optimal motion vector predictor, further decoding is calculated goes out the current block motion vector.

The 5th purpose of the embodiment of the invention is to propose a kind of motion-vector prediction coding/decoding system,, described system comprises described motion-vector prediction code device and described motion-vector prediction decoding device.

Beneficial effect of the present invention

The embodiment of the invention is by making up a cover motion vector analog codec module at coding side, select the predicted value of motion vector more accurately by the room and time correlation, reduce the code check of encoding motion vector difference, the compressing motion vector data reduce the code check of exporting more subtly.

Description of drawings

Fig. 1 is a kind of motion-vector prediction coding method of embodiment of the invention flow chart;

Fig. 2 carries out analog codec at coding side to motion vector in Fig. 1 method, to obtain the method flow diagram of the optimal motion vector analogue value;

Fig. 3 is that the current block enumerated in the embodiment of the invention is positioned at the 1st of present frame and is listed as the candidate motion vector location drawing of k when capable;

Fig. 4 is a kind of motion-vector prediction code device of embodiment of the invention structural representation;

Fig. 5 is the concrete modular structure schematic diagram of the analog codec module in Fig. 4 device;

Fig. 6 is a kind of motion-vector prediction coding/decoding method of embodiment of the invention flow chart;

Fig. 7 is a kind of motion-vector prediction decoding device of embodiment of the invention structural representation;

Fig. 8 is a kind of motion-vector prediction coding/decoding system of embodiment of the invention structural representation.

Embodiment

In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated, for convenience of explanation, only show the part relevant with the embodiment of the invention.Should be appreciated that the specific embodiment that this place is described, only be used to explain the present invention, not in order to restriction the present invention.

The embodiment of the invention is by making up a cover motion vector analog codec module at coding side, select the predicted value of motion vector more accurately by the room and time correlation, reduce the code check of encoding motion vector difference, the compressing motion vector data reduce the code check of exporting more subtly.

Embodiment one

Fig. 1 is a kind of motion-vector prediction coding method of embodiment of the invention flow chart, said method comprising the steps of:

S101 obtains the motion vector mv through current block behind the motion prediction;

S102 makes up the set of candidate motion vectors that is used for the motion-vector prediction coding;

Described " structure is used for the set of candidate motion vectors of motion-vector prediction coding " concrete grammar is: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors

[mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, for example: mv _APosition number in set of candidate motion vectors can be 1, mv _DPosition number in set of candidate motion vectors can be 5.

Being illustrated in figure 3 as the current block of enumerating is positioned at the 1st of present frame and is listed as the candidate motion vector location drawing of k when capable.As shown in Figure 3, the motion vector of current block is mv, and the motion vector of current block left side block is mv _A, the motion vector of piece is mv directly over the current block _B, the motion vector of the upper right square of current block is mv _CThe motion vector of present frame former frame same position (the 1st be listed as k capable) piece is mv _D, the motion vector of the reference block that current block motion vector mv points to is mv _E

S103 chooses optimal motion vector predictor mv from set of candidate motion vectors _Best

Describedly " choose optimal motion vector predictor mv _Best" concrete grammar as follows:

S1031, the bit number R (mv-mv that calculation code motion-vector prediction difference needs _P);

Wherein, mv _P∈ [mv _AMv _BMv _CMv _H264Mv _DMv _E], wherein mv is the current block motion vector, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select.

S1032 selects to make encoding motion vector difference bit number R (mv-mv _P) minimum motion vector predictor mv _PBe optimal motion vector predictor mv _Best

S104 asks for current block motion vector mv and described optimal motion vector predictor mv _BestMotion vector difference mvd, mvd=mv-mv _Best

S106 is sent to decoding end with described motion vector difference mvd through the entropy coding module coding.

The embodiment of the invention is selected the predicted value of motion vector more accurately by the room and time correlation, has reduced the code check of encoding motion vector difference.

In order to reduce the code check that transmitting moving vector prediction pattern needs further, after step S104, also can increase following steps:

S105 carries out analog codec at coding side to motion vector, to obtain the optimal motion vector analogue value;

Fig. 2 is that described coding side carries out analog codec to motion vector, to obtain the method flow diagram of the optimal motion vector analogue value;

Described " motion vector being carried out analog codec, to obtain the optimal motion vector analogue value " concrete grammar is:

S1051 selects motion vector predictor mv successively from described set of candidate motion vectors _P

S1052 makes up interim motion vector mvi, mvi=mvd+mvp;

S1053 obtains current reconstructed image blocks of data Y _R, Y _R=Z _R+ X _R

Wherein:

Z _RBe the prediction data of current block image, it carries out motion compensation by described interim motion vector mvi and obtains;

X _RBe residual error data, it carries out motion compensation by current block motion vector mv and obtains.

S1054 makes up coefficient of variation E _d

$E_d=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{[y(i,-1)-y(i,0)]}^2+\underset{j=1}{\overset{m-1}{\mathrm{\Σ}}}{[y(-1,j)-y(0,j)]}^2$

Wherein,

(i j) is the reconstructed image data of the capable j row of i, current reconstructed image piece to y

Data Y _RIn the upper left corner first data be initial point, with y (0,0) expression, m and n are the length of current block and wide;

Y (i ,-1) represents current reconstructed image blocks of data Y _RThe left side one columns certificate, wherein the value of i from 0 to n-1;

Y (1, j) Biao Shi current reconstructed image blocks of data Y _RThe top data line, wherein the value of j from 1 to m-1;

S1055 selects to make coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec

If described optimal motion vector analogue value mv _DecEqual optimal motion vector predictor mv _Dest, at known mvd and mv _BestSituation under, decoding end can be directly according to formula mvd=mv-mv in when decoding _BestDraw the motion vector mv of current block, then coding side need not increase the description that indicates the position number of optimal motion vector predictor in set of candidate motion vectors in output code flow;

If described optimal motion vector analogue value mv _DecBe not equal to optimal motion vector predictor mv _Best, decoding end can't correctly be selected optimal motion vector predictor mv _Best, then coding side need increase in output code flow and indicates optimal motion vector predictor mv _BestThe description of the position number in set of candidate motion vectors; Decoding end is obtained optimal motion vector predictor mv according to described position number _BestAt known mvd and mv _BestSituation under, thereby further according to formula mvd=mv-mv _BestDraw the motion vector mv of current block.

The embodiment of the invention proposes a kind of motion-vector prediction coding method, select the predicted value of motion vector more accurately by the room and time correlation, reduce the code check of encoding motion vector difference, the compressing motion vector data reduce the code check of exporting more subtly.

Embodiment two

Fig. 4 is a kind of motion-vector prediction code device of embodiment of the invention structural representation, and described motion-vector prediction code device comprises:

The motion vector acquisition module is used to obtain the motion vector through current block behind the motion prediction;

Set of candidate motion vectors makes up module, is used to make up the set of candidate motion vectors that is used for the motion-vector prediction coding; Be specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors.

The optimal motion vector predictor is chosen module, is used for choosing the optimal motion vector predictor from set of candidate motion vectors, and it comprises motion-vector prediction difference bit number computing module, motion vector difference selection module;

Motion-vector prediction difference bit number computing module is used for the bit number R (mv-mv of calculation code motion-vector prediction difference needs _P), wherein, mv _P∈ [mv _AMv _BMv _CMv _H264Mv _DMv _E], wherein mv is the current block motion vector, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select;

The motion vector difference bit number is selected module, is used for selecting to make encoding motion vector difference bit number R (mv-mv _P) minimum motion vector predictor mv _PBe optimal motion vector predictor mv _Best

Motion vector difference is asked for module, is used to ask for current block motion vector mv and described optimal motion vector predictor mv _BestMotion vector difference mvd, mvd=mv-mv _Best

Further, described motion-vector prediction code device also can comprise:

The motion vector difference sending module is used for described motion vector difference mvd is sent to decoding end through the entropy coding module coding;

Further, described motion-vector prediction code device can also comprise the analog codec module, is the concrete modular structure schematic diagram of described analog codec module as described in Figure 5.

The analog codec module is asked for module with described motion vector difference and is linked to each other, and is used at coding side motion vector being carried out analog codec, to obtain the optimal motion vector analogue value;

Described analog codec module comprises that further motion vector predictor is chosen module, motion vector structure module, reconstructed image blocks of data acquisition module, coefficient of variation make up module temporarily, the optimal motion vector analogue value is chosen module, optimal motion vector position sequence number writing module;

Motion vector predictor is chosen module, is used for selecting motion vector predictor mv from described set of candidate motion vectors successively _P

Interim motion vector makes up module, is used to make up interim motion vector mvi, mvi=mvd+mvp; Wherein mvd is the motion vector difference of current block motion vector and described optimal motion vector predictor, mvd=mv-mv _Best

Reconstructed image blocks of data acquisition module is used to obtain current reconstructed image blocks of data Y _R, Y _R=Z _R+ X _R

Wherein: Z _RBe the prediction data of current block image, it carries out motion compensation by described interim motion vector mvi and obtains;

X _RBe residual error data, it carries out motion compensation by current block motion vector mv and obtains.Coefficient of variation makes up module, is used to ask structure coefficient of variation E _d

$E_d=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{[y(i,-1)-y(i,0)]}^2+\underset{j=1}{\overset{m-1}{\mathrm{\Σ}}}{[y(-1,j)-y(0,j)]}^2$

Wherein,

(i j) is the reconstructed image data of the capable j row of i, current reconstructed image blocks of data Y to y _RIn the upper left corner first data be initial point, with y (0,0) expression, m and n are the length of current block and wide;

Y (i ,-1) represents current reconstructed image blocks of data Y _RThe left side one columns certificate, wherein the value of i from 0 to n-1;

Y (1, j) Biao Shi current reconstructed image blocks of data Y _RThe top data line, wherein the value of j from 1 to m-1;

The optimal motion vector analogue value is chosen module, is used for selecting to make coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec

Optimal motion vector position sequence number writing module is if be used for as described optimal motion vector analogue value mv _DecBe not equal to optimal motion vector predictor mv _Best, increase indicates optimal motion vector predictor mv in output code flow _BestThe description of the position number in set of candidate motion vectors.

The embodiment of the invention is selected the predicted value of motion vector more accurately by the room and time correlation, reduces the code check of encoding motion vector difference, and compressing motion vector data more subtly reduce the code check of output.

Embodiment three

Figure 6 shows that a kind of motion-vector prediction coding/decoding method of embodiment of the invention flow chart, described method comprises:

S601, received code end transmitted stream data;

S602 detects in the bit stream data that receives whether comprise the motion vector position number, is then to enter step S603, otherwise enters step S604

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors;

S603 is according to described motion vector position number, from the concentrated optimal motion vector predictor mv that obtains of motion vector candidates _Best, further decoding is calculated goes out current block motion vector mv;

Wherein, mvd=mv-mv _Best

S604, the motion vector difference mvd that receives decoded calculates the current block motion vector;

Wherein, mvd=mv-mv _Best

Described bit stream data comprises motion vector difference mvd and/or motion vector position number;

Described set of candidate motion vectors is made up by coding side, is specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, and described motion vector collection is safeguarded jointly by coding side and decoding end;

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors.

Embodiment four

Fig. 7 is a kind of motion-vector prediction decoding device of embodiment of the invention structural representation.

Described motion-vector prediction decoding device comprises data reception module, motion vector position number detection module, decoder module;

Data reception module is used for received code end transmitted stream data, and described bit stream data comprises motion vector difference and/or motion vector position number;

Motion vector position number detection module, whether the bit stream data that is used for detecting reception comprises the motion vector position number, and described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors;

Decoder module is used for when bit stream data does not comprise the position number of motion vector, and the motion vector difference that receives is decoded calculates the current block motion vector; Also be used for when bit stream data comprises the motion vector position number, according to described motion vector position number, concentrate from motion vector candidates and to obtain the optimal motion vector predictor, further decoding is calculated goes out the current block motion vector.

Described set of candidate motion vectors is made up by coding side, is specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, and described motion vector collection is safeguarded jointly by coding side and decoding end;

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors.

Implement five

Fig. 8 is a kind of motion-vector prediction coding/decoding system of embodiment of the invention structural representation, and described system comprises motion-vector prediction code device, motion-vector prediction decoding device.

Described motion-vector prediction code device comprises:

The motion vector acquisition module is used to obtain the motion vector through current block behind the motion prediction;

Set of candidate motion vectors makes up module, is used to make up the set of candidate motion vectors that is used for the motion-vector prediction coding; Be specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors;

The optimal motion vector predictor is chosen module, is used for choosing the optimal motion vector predictor from set of candidate motion vectors, and it comprises motion-vector prediction difference bit number computing module, motion vector difference selection module;

Motion-vector prediction difference bit number computing module is used for the bit number R (mv-mv of calculation code motion-vector prediction difference needs _P), wherein, mv _P∈ [mv _AMv _BMv _CMv _H264Mv _DMv _E], wherein mv is the current block motion vector, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select;

The motion vector difference bit number is selected module, is used for selecting to make encoding motion vector difference bit number R (mv-mv _P) minimum motion vector predictor mv _PBe optimal motion vector predictor mv _Best

Motion vector difference is asked for module, is used to ask for current block motion vector mv and described optimal motion vector predictor mv _BestMotion vector difference mvd, mvd=mv-mv _Best

Further, described motion-vector prediction code device also can comprise:

The motion vector difference sending module is used for described motion vector difference mvd is sent to decoding end through the entropy coding module coding;

Further, described motion-vector prediction code device can also comprise the analog codec module, is the concrete modular structure schematic diagram of described analog codec module as described in Figure 5.

The analog codec module is asked for module with described motion vector difference and is linked to each other, and is used at coding side motion vector being carried out analog codec, to obtain the optimal motion vector analogue value;

Described analog codec module comprises that further motion vector predictor is chosen module, motion vector structure module, reconstructed image blocks of data acquisition module, coefficient of variation make up module temporarily, the optimal motion vector analogue value is chosen module, optimal motion vector position sequence number writing module.

Motion vector predictor is chosen module, is used for selecting motion vector predictor mv from described set of candidate motion vectors successively _P

Interim motion vector makes up module, is used to make up interim motion vector mvi, mvi=mvd+mvp; Wherein mvd is the motion vector difference of current block motion vector and described optimal motion vector predictor, mvd=mv-mv _Best

Reconstructed image blocks of data acquisition module is used to obtain current reconstructed image blocks of data Y _R, Y _R=Z _R+ X _R

Wherein: Z _RBe the prediction data of current block image, it carries out motion compensation by described interim motion vector mvi and obtains;

X _RBe residual error data, it carries out motion compensation by current block motion vector mv and obtains.Coefficient of variation makes up module, is used to ask structure coefficient of variation E _d

$E_d=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{[y(i,-1)-y(i,0)]}^2+\underset{j=1}{\overset{m-1}{\mathrm{\Σ}}}{[y(-1,j)-y(0,j)]}^2$

Wherein,

(i j) is the reconstructed image data of the capable j row of i, current reconstructed image blocks of data Y to y _RIn the upper left corner first data be initial point, with y (0,0) expression, m and n are the length of current block and wide;

Y (i ,-1) represents current reconstructed image blocks of data Y _RThe left side one columns certificate, wherein the value of i from 0 to n-1;

Y (1, j) Biao Shi current reconstructed image blocks of data Y _RThe top data line, wherein the value of j from 1 to m-1;

The optimal motion vector analogue value is chosen module, is used for selecting to make coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec

Optimal motion vector position sequence number writing module is if be used for as described optimal motion vector analogue value mv _DecBe not equal to optimal motion vector predictor mv _Best, increase indicates optimal motion vector predictor mv in output code flow _BestThe description of the position number in set of candidate motion vectors.

Described motion-vector prediction decoding device comprises data reception module, motion vector position number detection module, decoder module;

Data reception module is used for received code end transmitted stream data, and described bit stream data comprises motion vector difference and/or motion vector position number;

Motion vector position number detection module, whether the bit stream data that is used for detecting reception comprises the motion vector position number, and described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors;

Decoder module is used for when bit stream data does not comprise the position number of motion vector, and the motion vector difference that receives is decoded calculates the current block motion vector; Also be used for when bit stream data comprises the motion vector position number, according to described motion vector position number, concentrate from motion vector candidates and to obtain the optimal motion vector predictor, further decoding is calculated goes out the current block motion vector.

Described set of candidate motion vectors is made up by coding side, is specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, and described motion vector collection is safeguarded jointly by coding side and decoding end;

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors.

The embodiment of the invention is selected the predicted value of motion vector more accurately by the room and time correlation, reduces the code check of encoding motion vector difference, and compressing motion vector data more subtly reduce the code check of output.

Those having ordinary skill in the art will appreciate that, all or part of step in realization the foregoing description method can be finished by the program command related hardware, described program can be stored in the computer read/write memory medium, and described storage medium can be ROM, RAM, disk, CD etc.

The above only is preferred embodiment of the present invention, not in order to restriction the present invention, all any modifications of being done within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.

Claims (15)

1. motion-vector prediction coding method, described method comprises:

Obtain motion vector (mv) through current block behind the motion prediction;

Structure is used for the set of candidate motion vectors of motion-vector prediction coding;

From set of candidate motion vectors, choose optimal motion vector predictor (mv _Best);

Ask for the motion vector difference (mvd) of current block motion vector and described optimal motion vector predictor;

To be sent to decoding end behind described motion vector difference (mvd) entropy coding;

It is characterized in that,

Described " structure is used for the set of candidate motion vectors of motion-vector prediction coding " concrete grammar is: the motion vector (mv that chooses the contiguous piece in the current block left side _A), the motion vector (mv of piece directly over the current block _B), the motion vector (mv of the upper right square of current block _C), the current block motion vectors (mv that uses the H264 method to obtain _H264), the motion vector (mv of present frame former frame same position piece _D), the motion vector (mv of the reference block of current block motion vector points _E), constitute set of candidate motion vectors ([mv _AMv _BMv _CMv _H264Mv _DMv _E]).

2. motion-vector prediction coding method as claimed in claim 1 is characterized in that, the motion vector difference that step is asked for current block motion vector and described optimal motion vector predictor also comprises step afterwards:

At coding side motion vector is carried out analog codec, to obtain the optimal motion vector analogue value (mv _Dec).

3. motion-vector prediction coding method as claimed in claim 1 is characterized in that, the described optimal motion vector predictor of choosing from set of candidate motion vectors is specially:

Bit number (R (the mv-mv that calculation code motion-vector prediction difference needs _P));

Selection makes encoding motion vector difference bit number (R (mv-mv _P)) minimum motion vector predictor (mv _P) be optimal motion vector predictor (mv _Best);

Wherein, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select, mv is the current block motion vector.

4. motion-vector prediction coding method as claimed in claim 2 is characterized in that, described " at coding side motion vector being carried out analog codec, to obtain the optimal motion vector analogue value " is specially:

From described set of candidate motion vectors, select motion vector predictor mv successively _P

Make up interim motion vector mvi, mvi=mvd+mvp;

Obtain current reconstructed image blocks of data Y _R, Y _R=Z _R+ X _R

Wherein: Z _RBe the prediction data of current block image, it carries out motion compensation by described interim motion vector mvi and obtains; X _RBe residual error data, it carries out motion compensation by current block motion vector mv and obtains;

Make up coefficient of variation E _d

$E_d=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{[y(i,-1)-y(i,0)]}^2+\underset{j=1}{\overset{m-1}{\mathrm{\Σ}}}{[y(-1,j)-y(0,j)]}^2$

Wherein,

(i j) is the reconstructed image data of the capable j row of i, current reconstructed image blocks of data Y to y _RIn the upper left corner first data be initial point, with y (0,0) expression, m and n are the length of current block and wide;

Y (i ,-1) represents current reconstructed image blocks of data Y _RThe left side one columns certificate, wherein the value of i from 0 to n-1;

Y (1, j) Biao Shi current reconstructed image blocks of data Y _RThe top data line, wherein the value of j from 1 to m-1;

Selection makes coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec

5. motion-vector prediction coding method as claimed in claim 4 is characterized in that, described " selection makes coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec" also comprise:

If described optimal motion vector analogue value mv _DecEqual optimal motion vector predictor mv _Best, then coding side need not increase the description that indicates the position number of optimal motion vector predictor in set of candidate motion vectors in output code flow;

If described optimal motion vector analogue value mv _DecBe not equal to optimal motion vector predictor mv _Best, then coding side need increase in output code flow and indicates optimal motion vector predictor mv _BestThe description mv of the position number in set of candidate motion vectors.

6. a motion-vector prediction code device is characterized in that, described device comprises:

The motion vector acquisition module is used to obtain the motion vector through current block behind the motion prediction;

Set of candidate motion vectors makes up module, is used to make up the set of candidate motion vectors that is used for the motion-vector prediction coding; Be specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors;

The optimal motion vector predictor is chosen module, is used for choosing the optimal motion vector predictor from set of candidate motion vectors, and it comprises motion-vector prediction difference bit number computing module, motion vector difference selection module;

Motion-vector prediction difference bit number computing module is used for the bit number R (mv-mv of calculation code motion-vector prediction difference needs _P), wherein, mv _P∈ [mv _AMv _BMv _CMv _H264Mv _DMv _E], wherein mv is the current block motion vector, mv _PBe the predicted value of motion vector, mv _PValue in described set of candidate motion vectors, select;

The motion vector difference bit number is selected module, is used for selecting to make encoding motion vector difference bit number R (mv-mv _P) minimum motion vector predictor mv _PBe optimal motion vector predictor mv _Best

Motion vector difference is asked for module, is used to ask for current block motion vector mv and described optimal motion vector predictor mv _BestMotion vector difference mvd, mvd=mv-mv _Best

7. motion-vector prediction code device as claimed in claim 6 is characterized in that, described device further comprises:

The motion vector difference sending module is used for described motion vector difference mvd is sent to decoding end through the entropy coding module coding.

8. motion-vector prediction code device as claimed in claim 6 is characterized in that, described device further comprises:

The analog codec module is asked for module with described motion vector difference and is linked to each other, and is used at coding side motion vector being carried out analog codec, to obtain the optimal motion vector analogue value.

9. motion-vector prediction code device as claimed in claim 8 is characterized in that, described analog codec module comprises:

Motion vector predictor is chosen module, is used for selecting motion vector predictor mv from described set of candidate motion vectors successively _P

Interim motion vector makes up module, is used to make up interim motion vector mvi, mvi=mvd+mvp; Wherein mvd is the motion vector difference of current block motion vector and described optimal motion vector predictor, mvd=mv-mv _Best

Reconstructed image blocks of data acquisition module is used to obtain current reconstructed image blocks of data Y _R, Y _R=Z _R+ X _R,

Wherein: Z _RBe the prediction data of current block image, it carries out motion compensation by described interim motion vector mvi and obtains,

X _RBe residual error data, it carries out motion compensation by current block motion vector mv and obtains; Coefficient of variation makes up module, is used to ask structure coefficient of variation E _d,

$E_d=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{[y(i,-1)-y(i,0)]}^2+\underset{j=1}{\overset{m-1}{\mathrm{\Σ}}}{[y(-1,j)-y(0,j)]}^2$

Wherein,

(i j) is the reconstructed image data of the capable j row of i, current reconstructed image blocks of data Y to y _RIn the upper left corner first data be initial point, with y (0,0) expression, m and n are the length of current block and wide,

Y (i ,-1) represents current reconstructed image blocks of data Y _RThe left side one columns certificate, wherein the value of i is from 0 to n-1,

Y (1, j) Biao Shi current reconstructed image blocks of data Y _RThe top data line, wherein the value of j from 1 to m-1;

The optimal motion vector analogue value is chosen module, is used for selecting to make coefficient of variation E _dMinimum motion vector predictor mv _PBe the decoding optimal motion vector analogue value mv that simulates _Dec

Optimal motion vector position sequence number writing module is if be used for as described optimal motion vector analogue value mv _DecBe not equal to optimal motion vector predictor mv _Best, increase indicates optimal motion vector predictor mv in output code flow _BestThe description of the position number in set of candidate motion vectors.

10. a motion-vector prediction coding/decoding method is characterized in that, described method comprises:

Received code end transmitted stream data;

Detect in the bit stream data that receives and whether comprise the position number of motion vector in set of candidate motion vectors;

According to whether comprising the motion vector position number in the bit stream data, decoding in different ways.

" according to whether comprising the motion vector position number in the bit stream data, adopting different modes to decode " is specially 11. motion-vector prediction coding/decoding method as claimed in claim 10 is characterized in that, described:

When comprising described position number in the described bit stream data, according to described motion vector position number, from the concentrated optimal motion vector predictor mv that obtains of motion vector candidates _Best, further decoding is calculated goes out current block motion vector mv, wherein, and mvd=mv-mv _Best

When not comprising described position number in the described bit stream data, the motion vector difference mvd that receives decoded calculates the current block motion vector, wherein, and mvd=mv-mv _Best

12. as claim 10 or 11 described motion-vector prediction coding/decoding methods, it is characterized in that,

Described bit stream data comprises motion vector difference mvd and/or motion vector position number;

Described set of candidate motion vectors is made up by coding side, is specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, and described motion vector collection is safeguarded jointly by coding side and decoding end;

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors.

13. a motion-vector prediction decoding device is characterized in that, described device comprises:

Data reception module is used for received code end transmitted stream data, and described bit stream data comprises motion vector difference and/or motion vector position number;

Motion vector position number detection module, whether the bit stream data that is used for detecting reception comprises the motion vector position number, and described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors;

Decoder module is used for when bit stream data does not comprise the position number of motion vector, and the motion vector difference that receives is decoded calculates the current block motion vector; Also be used for when bit stream data comprises the motion vector position number, according to described motion vector position number, concentrate from motion vector candidates and to obtain the optimal motion vector predictor, further decoding is calculated goes out the current block motion vector.

14. motion-vector prediction decoding device as claimed in claim 13 is characterized in that,

Described set of candidate motion vectors is made up by coding side, is specially: the motion vector mv that chooses the contiguous piece in the current block left side _A, the motion vector mv of piece directly over the current block _B, the upper right square of current block motion vector mv _C, the current block motion vectors mv that uses the H264 method to obtain _H264, present frame former frame same position piece motion vector mv _D, the reference block that points to of current block motion vector mv motion vector mv _E, constitute set of candidate motion vectors [mv _AMv _BMv _CMv _H264Mv _DMv _E], each candidate motion vector all has its certain location sequence number in set of candidate motion vectors, and described motion vector collection is safeguarded jointly by coding side and decoding end;

Described motion vector position number is the certain location sequence number of candidate motion vector in set of candidate motion vectors.

15. a motion-vector prediction coding/decoding system is characterized in that, described system comprises claim 6 and described motion-vector prediction code device of claim 13 and motion-vector prediction decoding device.

Images