[go: up one dir, main page]

CN103686166B - Fast prediction mode selection method and system based on correlation analysis - Google Patents

Fast prediction mode selection method and system based on correlation analysis Download PDF

Info

Publication number
CN103686166B
CN103686166B CN201310574825.XA CN201310574825A CN103686166B CN 103686166 B CN103686166 B CN 103686166B CN 201310574825 A CN201310574825 A CN 201310574825A CN 103686166 B CN103686166 B CN 103686166B
Authority
CN
China
Prior art keywords
mode
present encoding
prediction
block
predictive mode
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201310574825.XA
Other languages
Chinese (zh)
Other versions
CN103686166A (en
Inventor
舒倩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
SHENZHEN YUNZHOU MULTIMEDIA TECHNOLOGY Co Ltd
Original Assignee
SHENZHEN YUNZHOU MULTIMEDIA TECHNOLOGY Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by SHENZHEN YUNZHOU MULTIMEDIA TECHNOLOGY Co Ltd filed Critical SHENZHEN YUNZHOU MULTIMEDIA TECHNOLOGY Co Ltd
Priority to CN201310574825.XA priority Critical patent/CN103686166B/en
Publication of CN103686166A publication Critical patent/CN103686166A/en
Application granted granted Critical
Publication of CN103686166B publication Critical patent/CN103686166B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Compression, Expansion, Code Conversion, And Decoders (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

The invention provides a fast prediction mode selection method and system based on correlation analysis. The method of the invention, according to correlation between boundary pixels within a block and surrounding blocks, deletes a prediction mode with little probability of occurrence or directly determines a best prediction mode, so that the calculation amount of the prediction mode traversal optimization is reduced while maintaining the stability of the rate distortion performance.

Description

A kind of predictive mode fast selecting method and system based on correlation analysis
Technical field
The present invention relates to coding and decoding video field, more particularly to a kind of predictive mode based on correlation analysis is quickly selected Method and system.
Background technology
In conventional intraframe prediction algorithm, a block generally requires to carry out traversal optimizing on various predictive modes, to obtain Take optimum prediction mode.The only block of 4x4 just has 9 kinds of predictive modes in H264, and the infra-frame prediction performance boost hand of main flow at present Section is that by taking the H265 for being in development as an example, its intra prediction mode is at maximum up to 34 kinds by increasing predictive mode.Now, Travel through according to conventional predictive mode, although the optimization realization of distortion performance, the huge calculating for bringing therewith can be reached Amount will restrict its application in Practical Project.
The content of the invention
The purpose of the embodiment of the present invention is to propose a kind of predictive mode fast selecting method based on correlation analysis, purport Solving the problems, such as that the conventional predictive mode traversal amount of calculation of prior art is huge.
The embodiment of the present invention is achieved in that a kind of quick selecting party of intra prediction mode based on correlation analysis Method, the fast intra-frame predicting method based on correlation analysis specifically includes following steps:
S1061:Calculate the directional correlation coefficient of present encoding block;
S1062:According to the different situations of the directional correlation coefficient of present encoding block, corresponding intra prediction mode is selected to make For optimal intra prediction mode Candidate Set;
S1063:The infra-frame prediction of the optimal intra prediction mode Candidate Set correspondence direction is carried out to present encoding block.
Second purpose of the embodiment of the present invention is to propose that a kind of intra prediction mode based on correlation analysis is quickly selected System is selected, the system includes that directional correlation coefficients calculation block, optimal intra prediction mode candidate select module, optimum frame Interior prediction module,
Directional correlation coefficients calculation block, for calculating the directional correlation coefficient of present encoding block;
Optimal intra prediction mode candidate selects module, for not sympathizing with according to the directional correlation coefficient of present encoding block Condition, selects corresponding intra prediction mode as optimal intra prediction mode Candidate Set;
Optimal intra-framed prediction module, for carrying out optimal intra prediction mode Candidate Set correspondence direction to present encoding block Infra-frame prediction.
3rd purpose of the embodiment of the present invention is to propose a kind of quick selecting party of predictive mode based on correlation analysis Method, the method comprising the steps of,
S101:Judge whether present frame is inter-frame, if yes then enter S102;
S102:The first predictive mode is carried out to present encoding basic block quickly to judge;
S103c:If the quick judged result of the first predictive mode is intra prediction mode, step S104 is directly entered;
S104:Calculate the block complexity of present encoding basic block;
S105:Judge whether mbit,n>Thres1, S106 is if so, then entered, if it is not, S107 is then entered, wherein, mbit,n Represent the block complexity of present encoding basic block, Thres1Represent the first decision threshold;
S106:Present encoding basic block is divided into into the encoding block less than basic block size, to the encoding block using such as The Adaptive Mode Selection Method for Intra-Prediction based on correlation analysis described in claim 1 carries out infra-frame prediction;
S107:Using the Adaptive Mode Selection Method for Intra-Prediction pair based on correlation analysis as claimed in claim 1 Present encoding basic block carries out infra-frame prediction;
S108:RDO is carried out to present encoding basic block, optimum prediction mode is obtained;
S109:Change quantization, scanning are carried out to present encoding basic block, entropy code operation completes present encoding basic block Coding.
S110:Judge that next coding basic block whether there is, if it is present next coding basic block is set to currently Encoding block, reenters S102;Otherwise judge that next coded frame whether there is, if it is present next coded frame is set to Current encoded frame, reenters S101;Otherwise, terminate.
It is further, further comprising the steps of after step S102,
S103d:If the quick judged result of the first predictive mode be predictive mode cannot determinating mode, to present encoding Basic block carries out inter prediction, subsequently into step S104.
It is further, further comprising the steps of after step S102,
S103a:If the quick judged result of the first predictive mode is Skip patterns, present encoding basic block is entered first Row Skip is predicted, subsequently into S109.
It is further, further comprising the steps of after step S102,
S103b:If the quick judged result of the first predictive mode is inter-frame forecast mode, basic to present encoding first Block carries out inter prediction, subsequently into S108.
Further, in step S101,
Judge whether present frame is inter-frame, if it is not, then into step S104.
4th purpose of the embodiment of the present invention is to propose that a kind of predictive mode based on correlation analysis quickly selects to be System, the system includes:
Inter-frame judge module, it is fast if yes then enter the first predictive mode for judging whether present frame is inter-frame Fast judge module, if otherwise entering block complicated dynamic behaviour module;;
The quick judge module of first predictive mode, quickly sentences for carrying out the first predictive mode to present encoding basic block It is disconnected;
First judges processing meanss, for according to the quick judged result of the first predictive mode, entering to present encoding basic block Row is processed;
Block complicated dynamic behaviour module, for calculating the block complexity of present encoding basic block;
5th judge module, for judging whether mbit,n>Thres1, if so, then enter the first fast intra-mode prediction mould Block;If it is not, then entering the second fast intra-mode prediction module;
mbit,nRepresent the block complexity of present encoding basic block, Thres1Represent the first decision threshold, general Thres1> 25;
First fast intra-mode prediction module, for present encoding basic block to be divided into the coding less than basic block size Block, it is then quickly square using the intra prediction mode based on correlation analysis as claimed in claim 1 to each encoding block Method carries out infra-frame prediction;
Second fast intra-mode prediction module, for using the frame based on correlation analysis as claimed in claim 1 Inner estimation mode fast method carries out infra-frame prediction to present encoding basic block.
Optimum prediction mode acquisition module, judges that processing meanss and next code module are connected, for current with first Coding basic block carries out RDO, obtains optimum prediction mode;
Next code module, judges that processing meanss and next coding basic block judging treatmenting module are connected with first, is used for Change quantization, scanning are carried out to present encoding basic block, entropy code operation completes the coding of present encoding basic block;
Next coding basic block judging treatmenting module, for judging that next coding basic block whether there is, if it is present Next coding basic block is set to into present encoding block, the quick judge module of the first predictive mode is reentered;Otherwise, under entrance One coded frame judging treatmenting module;
Next coded frame judging treatmenting module, for judging that next coded frame whether there is, if it is present by next volume Code frame is set to current encoded frame, reenters inter-frame judge module;Otherwise, terminate.
Beneficial effects of the present invention
The present invention proposes a kind of predictive mode fast selecting method based on correlation analysis, according to block inner boundary pixel with The correlation of surrounding block, deletes the little predictive mode of probability of occurrence or directly determines optimal prediction modes, so as to reach reduction Predictive mode travels through the amount of calculation of optimizing, while keeping the stability of distortion performance.
Description of the drawings
Fig. 1 is Adaptive Mode Selection Method for Intra-Prediction flow chart of the preferred embodiment of the present invention based on correlation analysis;
Fig. 2 is a kind of predictive mode fast selecting method flow chart based on correlation analysis of the preferred embodiment of the present invention;
Fig. 3 is the 4x4 intra prediction mode and its prediction direction figure of preferred embodiment of the present invention h264;
Fig. 4 is that a kind of intra prediction mode based on correlation analysis of the preferred embodiment of the present invention quickly selects system architecture Figure;
Fig. 5 is that a kind of predictive mode based on correlation analysis of the preferred embodiment of the present invention quickly selects system construction drawing;
Fig. 6 is the structure chart of the quick judge module of the first predictive mode in Fig. 5;
Fig. 7 is the first structure chart for judging processing meanss in Fig. 5;
Fig. 8 is the structure chart of the first fast intra-mode prediction module in Fig. 5;
Fig. 9 is the structure chart of block complicated dynamic behaviour module in Fig. 5.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, it is right below in conjunction with drawings and Examples The present invention is further elaborated, and for convenience of description, illustrate only the part related to the embodiment of the present invention.Should manage Solution, the specific embodiment that this place is described is used only for explaining the present invention, not to limit the present invention.
The embodiment of the present invention proposes a kind of predictive mode fast selecting method based on correlation analysis, according to block inner boundary Pixel and the correlation of surrounding block, delete the little predictive mode of probability of occurrence or directly determine optimal prediction modes, so as to reach The amount of calculation of optimizing is traveled through to predictive mode is reduced, while keeping the stability of distortion performance.
Embodiment one
Fig. 1 is a kind of Adaptive Mode Selection Method for Intra-Prediction flow process based on correlation analysis of the preferred embodiment of the present invention Figure, methods described includes:
S1061:Calculate the directional correlation coefficient of present encoding block.
(A)If the predictive mode support of present encoding block and only the 4x4 intra prediction modes of support h264 is all pre- Direction is surveyed, then calculates the vertical of present encoding block, level, right diagonal, left diagonally opposed coefficient correlation, its computational methods is as follows: (H265 is included)
Wherein
Wherein, cov, std, mean (variable | condition)Represent respectively and seek covariance, ask mean square deviation, the change to meeting condition Amount is averaged;YbRepresent the brightness value of present encoding block;Yb(i1,:) represent the i-th of present encoding block1Row brightness value;Yb(i2,:) Represent the i-th of present encoding block2Row brightness value;Yb(:,j1) represent present encoding block jth1Row row brightness value;Yb(:,j2) table Show the jth of present encoding block2Row row brightness value;Yb(i, j | condition) represent what the i-th row of the present encoding block jth for meeting condition was arranged Brightness value;Size (b) represents present encoding block row size;rV、rH、rDDR、rDDLRespectively represent the vertical of present encoding block, level, Right diagonal, left diagonally opposed coefficient correlation, its subscript represents respectively corresponding prediction direction;rV,i1,i2、rH,j1,j2、 rDDR,k1,k2、rDDL,k1,k2, respectively represent vertical, level, right diagonal, left diagonally opposed coefficient correlation intermediate variable;k1、k2 First, second ranks and difference variable are represented respectively.V, H, DDL, DDR, VL, VR, HU, HD represent the 4x4 infra-frame prediction moulds of h264 Prediction direction title, is also corresponding predictive mode title.
(B)If the predictive mode of present encoding block is only supported in the 4x4 intra prediction modes of h264
A part of direction, then the corresponding related direction coefficient correlation of the predictive mode do not supported without calculate,
Delete relevant variable therein in corresponding computing formula.
The infra-frame prediction of the 16x16 of such as h264 only has tetra- kinds of predictive modes of V, H, DC, Plan(V, H, DC, Plan, expression The 16x16 infra-frame prediction mould prediction direction titles of h264, are also corresponding predictive mode title), now need to only calculate rV、rH, when Vertical, horizontal direction coefficient correlation the computational methods of front encoding block are as follows:
(C)If the predictive mode of present encoding block is in addition to prediction direction in the 4x4 intra prediction modes for supporting h264, The prediction direction more segmented also is supported, then optional selection method one or method two.
Method one:Using step(A)Middle process step.
Method two:Calculation procedure(A)Middle directional correlation coefficient and the coefficient correlation in subdivision direction.
Wherein, method can be according to step " to segment the coefficient correlation in direction "(A)In method directional correlation coefficient is done into class Push away.
The computational methods of subdivision directional correlation coefficient are exemplified below, the predictive mode of such as encoding block is except comprising HD Outward, also comprising the prediction direction between DDR and HD, comprising the prediction direction between HD and H, i.e., there is subdivision prediction in HD directions Pattern.With according to step(A)Similar approach calculated direction correlation coefficient rHD(Fig. 3 is the 4x4 intra prediction modes of h264 and its pre- Directional diagram is surveyed, V, H, DDL, DDR, VL, VR, HU, HD represent direction title),
,
Wherein
rHDThe coefficient correlation in HD directions is represented, its subscript represents respectively corresponding prediction direction, rHD,k1,k2Represent HD directions Coefficient correlation intermediate variable;k1、k2First, second ranks and difference variable are represented respectively.
S1062:According to the different situations of the directional correlation coefficient of present encoding block, corresponding intra prediction mode is selected to make For optimal intra prediction mode Candidate Set.
(a)If the 4x4 infra-frame prediction moulds of h264 are supported and only supported to the predictive mode of present encoding block
All prediction directions of formula, then
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL) corresponding predictive mode, max less (rV,rH,rDDR,rDDL) corresponding prediction mould Formula and between above-mentioned two prediction direction corresponding predictive mode, directionless predictive mode, coefficient correlation be equal to 2 pairs The predictive mode answered }
Situation 3:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL) corresponding predictive mode, positioned at above-mentioned prediction direction and max less (rV,rH, rDDR,rDDL) correspondingly corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictions between prediction direction Pattern }
Situation 4:If
Then optimal intra prediction mode Candidate Set
={max less(rV,rH,rDDR,rDDL) corresponding predictive mode, positioned at above-mentioned prediction direction and max (rV,rH, rDDR,rDDL) correspondingly corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictions between prediction direction Pattern }
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL) correspond to prediction direction and max (rV,rH,rDDR,rDDL) correspondence prediction Corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes between direction }
Situation 6:If above-mentioned situation is all unsatisfactory for, optimal intra prediction mode Candidate Set={ all predictive modes }.
Wherein, max(Variable)The maximum for being not equal to 2 variables is sought in expression;maxless(Variable)Expression is asked and is not equal to 2 changes The secondary maximum of amount;Abs is represented and is sought absolute value;rinv_maxRepresent perpendicular to max (rV,rH,rDDR,rDDL) correspond in prediction direction Corresponding coefficient correlation;rinv_maxlessRepresent perpendicular to max less (rV,rH,rDDR,rDDL) correspondence prediction direction on it is corresponding Coefficient correlation;Thres2Represent the second decision threshold, Thres3Represent the 3rd decision threshold, Thres4The 4th decision threshold is represented, Thres5Represent the general Thres of the 5th decision threshold2>0.95, general Thres4<0.3<Thres5<0.5<Thres3;It is " directionless Predictive mode " refers to the predictive mode that cannot sort out prediction direction, the DC predictive modes of such as h264, Plan predictive modes.
(b)If the predictive mode of present encoding block only supports a part of direction in the 4x4 intra prediction modes of h264, Then according to step(a)In method relevant variable is removed in " optimal intra prediction mode Candidate Set " calculation procedure.
The infra-frame prediction of the 16x16 of such as h264 only has tetra- kinds of predictive modes of V, H, DC, Plan, i.e., only support h264's Tri- directions of V, H, DC in 4x4 intra prediction modes, now only need to be according to step(a)In method in " optimal infra-frame prediction Relevant variable is removed in mode candidate collection " calculation procedure, it is specific as follows:
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH) corresponding predictive mode, max less (rV,rH) corresponding predictive mode and positioned at above-mentioned Corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes between two prediction directions }
Situation 3:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH) corresponding predictive mode, positioned at above-mentioned prediction direction and max less (rV,rH) correspondence prediction side Corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes between }
Situation 4:If
Then optimal intra prediction mode Candidate Set
={max less(rV,rH) corresponding predictive mode, positioned at above-mentioned prediction direction and max (rV,rH) correspondence prediction side Corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes between }
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL) correspond to prediction direction and max (rV,rH,rDDR,rDDL) correspondence prediction Corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes between direction }
Situation 6:If above-mentioned situation is all unsatisfactory for, optimal intra prediction mode Candidate Set={ all predictive modes }.
In this case, rinv_maxRepresent perpendicular to max (rV,rH) correspond to corresponding coefficient correlation in prediction direction; rinv_maxlessRepresent perpendicular to max less (rV,rH) correspond to corresponding coefficient correlation in prediction direction.
(c)If the predictive mode of present encoding block is in addition to prediction direction in the 4x4 intra prediction modes for supporting h264, The prediction direction more segmented also is supported, then following methods three or method four may be selected.
Method three:Using step(a)Middle process step.
Method four:According to step(a)In method increase phase in " optimal intra prediction mode Candidate Set " calculation procedure Dependent variable.
When the prediction direction for supporting more to segment is exemplified below, the determination side of its optimal intra prediction mode Candidate Set Method.The predictive mode of such as encoding block except comprising in addition to HD, also comprising the prediction direction between DDR and HD, comprising between HD and H Prediction direction, i.e., HD directions exist subdivision predictive mode.Now, according to step(a)In method in " optimal infra-frame prediction Increase relevant variable in mode candidate collection " calculation procedure, it is specific as follows
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfThen Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, max less (rV,rH,rDDR,rDDL,rHD) corresponding Predictive mode and between above-mentioned two prediction direction corresponding predictive mode, directionless predictive mode, coefficient correlation etc. In 2 corresponding predictive modes }
Situation 3:IfThen Optimal intra prediction mode Candidate Set
={max(rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, positioned at above-mentioned prediction direction and max less (rV, rH,rDDR,rDDL,rHD) correspondingly corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 correspondences between prediction direction Predictive mode
Situation 4:If
Then optimal intra prediction mode Candidate Set
={max less(rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, positioned at above-mentioned prediction direction and max (rV, rH,rDDR,rDDL,rHD) correspondingly corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 correspondences between prediction direction Predictive mode
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL,rHD) correspond to prediction direction and max (rV,rH,rDDR,rDDL,rHD) right Corresponding predictive mode between prediction direction, directionless predictive mode, coefficient correlation is answered to be equal to 2 corresponding predictive modes }
Situation 6:If above-mentioned situation is all unsatisfactory for, optimal intra prediction mode Candidate Set={ all predictive modes } exists In such cases, rinv_maxRepresent perpendicular to max (rV,rH,rDDR,rDDL,rHD) correspond to corresponding coefficient correlation in prediction direction; rinv_maxlessRepresent perpendicular to max less (rV,rH,rDDR,rDDL,rHD) correspond to corresponding coefficient correlation in prediction direction.
S1063:Present encoding block only carries out the infra-frame prediction of optimal intra prediction mode Candidate Set correspondence direction.
Embodiment two
Fig. 2 is a kind of predictive mode fast selecting method flow process based on correlation analysis of the preferred embodiment of the present invention Figure;The method comprising the steps of:
S101:Judge whether present frame is inter-frame, if yes then enter S102;Otherwise, into S104.
S102:The first predictive mode is carried out to present encoding basic block quickly to judge;
First predictive mode quick judgment method is as follows:
S1020:Calculate the symbolic variable flag of present encoding basic block;
S1021:Calculate the time correlation degree of present encoding basic block;
Yt,n={yt(i,j)|yt(i,j)∈mbt,n}、Yt-1,n={yt-1(i,j)|yt-1(i,j)∈mbt-1,n}
dYt,n={yt(i,j)-yt-1(i,j)|yt(i,j)∈mbt,nAnd yt-1(i,j)∈mbt-1,n}
Wherein, cov, std are represented respectively and are sought covariance, seek mean square deviation;yt(i, j) represents the i-th row of current encoded frame jth row Brightness value;yt-1(i, j) represents the brightness value of the i-th row of former frame jth row;mbt,nRepresent present encoding basic block;mbt-1,nTable Show that former frame same position encodes basic block;T represents image time sequence number in the video sequence;N represents basic block in its institute Position number in the picture;ρtt,nRepresent the time correlation degree of present encoding basic block;Basic block refers to the block that encoder is supported Full-size, such as basic block herein is exactly the macro block of 16x16 in H264, and can be for 64x64 etc. in H265 Large-sized piece;
S1022:If(ρtt,n>Const0 and flag=1)Or (ρ tt,n=0 and flag=0), then present encoding base This block is Skip patterns;
Otherwise, if(ρtt,n>Const1 and flag=1)Or (ρ tt,n<Const3 and flag=0), then currently compile Code basic block is inter-frame forecast mode;
Else if(ρtt,n<Const2 and flag=1)Or (ρ tt,n>Const4 and flag=0), it is current to compile basic Block is intra prediction mode,
Otherwise,(When above-mentioned condition is unsatisfactory for)The current predictive mode for compiling basic block cannot judge.
Wherein, const0, const1, const2, const3, const4 are respectively corresponding decision threshold;It is general desirable 0.99≤const0≤1,0.7<Const1≤1,0≤const2<0.3;Const3=f1 (fps), const4=f2 (fps), f1 For the monotonic decreasing function of frame per second, f2 for frame per second monotonic decreasing function, typically under full frame rate const3≤5, const4 >= 10;Fps represents frame per second.
S103:If the quick judged result of the first predictive mode is Skip patterns, present encoding basic block is entered first Row Skip is predicted, subsequently into S109;(This step is S103a)
If the quick judged result of the first predictive mode is inter-frame forecast mode, present encoding basic block is carried out first Inter prediction, subsequently into S108;(This step is S103b)
If the quick judged result of the first predictive mode is intra prediction mode, step S104 is directly entered;(This step For S103c)
If the quick judged result of the first predictive mode be predictive mode cannot determinating mode, to present encoding basic block Inter prediction is carried out, subsequently into step S104.(This step is S103d)
Step S103 includes step S103a, S103b, S103c, S103d, and step S103a, S103b, S103c, S103d are not Successively sequentially.
S104:Calculate the block complexity of present encoding basic block.Specific implementation is as follows
S1041:Calculate the boundary intensity of each pixel in present encoding basic block
yit(i,j)=‖filter(yt(i,j)|yt(i,j)∈mbt,n)‖
S1042:Mean square deviation is asked to the boundary intensity of each pixel in present encoding basic block
mbit,n=std(yit(i,j)|yt(i,j)∈mbt,n)
Wherein, yit(i, j) represents pixel ytThe boundary intensity of (i, j);Filter (variable | condition) represent to meeting condition Variable carry out boundary filtering, known any one boundary filter operator such as sobel may be selected;‖ variable ‖ are represented to variable Modulus;mbit,nRepresent the block complexity of present encoding basic block.
S105:Judge whether mbit,n>Thres1, if so, then enter S106;If it is not, then entering S107.
Wherein, Thres1Represent the first decision threshold, general Thres1>25。
S106:Present encoding basic block is divided into into the encoding block less than basic block size, to the encoding block using real Applying the fast intra-frame predicting method based on correlation analysis described in example one carries out infra-frame prediction;
For example in H264-baseline or H264-main, the quick frame in that step S106 as only carries out 4x4 is pre- Survey, and in step H264-high, S106 is the fast intra-mode prediction for carrying out 8x8,4x4.
S107:Using the fast intra-frame predicting method based on correlation analysis described in embodiment one to present encoding base This block carries out infra-frame prediction.
For example in H264, S107 is the fast intra-mode prediction for carrying out 16x16.
S108:RDO is carried out to present encoding basic block(Rate-Distortion Optimization), obtain optimal pre- Survey pattern.
S109:Normal coded stream journey subsequent operation is carried out to present encoding basic block;
Change quantization, scanning are carried out to present encoding basic block, entropy code operation completes the volume of present encoding basic block Code.
S110:Judge that next coding basic block whether there is, if it is present next coding basic block is set to currently Encoding block, reenters S102;Otherwise, it is determined that next coded frame whether there is, if it is present next coded frame is set to Current encoded frame, reenters S101;Otherwise, terminate.
Embodiment three
Fig. 4 is that a kind of intra prediction mode based on correlation analysis of the preferred embodiment of the present invention quickly selects system architecture Figure;The system includes:
Directional correlation coefficients calculation block, optimal intra prediction mode candidate select module, optimal intra-framed prediction module,
Directional correlation coefficients calculation block, for calculating the directional correlation coefficient of present encoding block;
The computational methods of the directional correlation coefficient of present encoding block are referring to the technical side in step S1061 in embodiment one Case, will not be described here.
Optimal intra prediction mode candidate selects module, for not sympathizing with according to the directional correlation coefficient of present encoding block Condition, selects corresponding intra prediction mode as optimal intra prediction mode Candidate Set;
It is described " to be used for the different situations of the directional correlation coefficient according to present encoding block, select corresponding infra-frame prediction mould Formula is used as optimal intra prediction mode Candidate Set " method detailed is referring to the technical scheme in step S1062 in embodiment one, here Repeat no more.
Optimal intra-framed prediction module, for carrying out optimal intra prediction mode Candidate Set correspondence direction to present encoding block Infra-frame prediction;
Example IV
Fig. 5 is that a kind of predictive mode based on correlation analysis of the preferred embodiment of the present invention quickly selects system construction drawing; The system includes:Inter-frame judge module, the quick judge module of the first predictive mode, first judge that processing meanss, block are complicated Degree computing module, the 5th judge module, the first fast intra-mode prediction module, the second fast intra-mode prediction module, optimum prediction mould Formula acquisition module, next code module, next coding basic block judging treatmenting module, next coded frame judging treatmenting module,
Inter-frame judge module, it is fast if yes then enter the first predictive mode for judging whether present frame is inter-frame Fast judge module, otherwise, into block complicated dynamic behaviour module;
The quick judge module of first predictive mode, quickly sentences for carrying out the first predictive mode to present encoding basic block It is disconnected;
First judges processing meanss, for according to the quick judged result of the first predictive mode, entering to present encoding basic block Row is processed;
Block complicated dynamic behaviour module, for calculating the block complexity of present encoding basic block;Block complicated dynamic behaviour module, bag Boundary strength calculation module, boundary intensity mean square deviation computing module are included, Fig. 9 is the structure chart of block complicated dynamic behaviour module in Fig. 5;
Boundary strength calculation module, for calculating the boundary intensity of each pixel in present encoding basic block;
yit(i,j)=‖filter(yt(i,j)|yt(i,j)∈mbt,n)‖
Boundary intensity mean square deviation computing module, for asking square to the boundary intensity of each pixel in present encoding basic block Difference;
mbit,n=std(yit(i,j)|yt(i,j)∈mbt,n)
Wherein, yit(i, j) represents pixel ytThe boundary intensity of (i, j);Filter (variable | condition) represent to meeting condition Variable carry out boundary filtering, known any one boundary filter operator such as sobel may be selected;‖ variable ‖ are represented to variable Modulus;mbit,nRepresent the block complexity of present encoding basic block.
5th judge module, for judging whether mbit,n>Thres1, if so, then enter the first fast intra-mode prediction mould Block;If it is not, then entering the second fast intra-mode prediction module;
mbit,nRepresent the block complexity of present encoding basic block, Thres1The first decision threshold is represented, typically
Thres1>25;
First fast intra-mode prediction module, for present encoding basic block to be divided into the coding less than basic block size Block, then to the encoding block using the quick selecting party of the intra prediction mode based on correlation analysis described in embodiment one Method carries out infra-frame prediction;
Second fast intra-mode prediction module, for using the infra-frame prediction based on correlation analysis described in embodiment one Mode quick selecting method carries out infra-frame prediction to present encoding basic block;
Optimum prediction mode acquisition module, for carrying out RDO to present encoding basic block(Rate-Distortion Optimization), obtain optimum prediction mode;
Next code module, for carrying out change quantization, scanning to present encoding basic block, entropy code operation completes to work as The coding of front coding basic block;
Next coding basic block judging treatmenting module, for judging that next coding basic block whether there is, if it is present Next coding basic block is set to into present encoding block, the quick judge module of the first predictive mode is reentered;Otherwise, under entrance One coded frame judging treatmenting module;
Next coded frame judging treatmenting module, for judging that next coded frame whether there is, if it is present by next volume Code frame is set to current encoded frame, reenters inter-frame judge module;Otherwise, terminate.
Further, the quick judge module of first predictive mode also includes symbolic variable computing module, time correlation Degree computing module, predictive mode judge module, Fig. 6 is the structure chart of the quick judge module of the first predictive mode in Fig. 5;
Symbolic variable computing module, for calculating the symbolic variable flag of present encoding basic block;
Time correlation degree computing module, for calculating the time correlation degree of present encoding basic block;
Yt,n={yt(i,j)|yt(i,j)∈mbt,n}、Yt-1,n={yt-1(i,j)|yt-1(i,j)∈mbt-1,n}
dYt,n={yt(i,j)-yt-1(i,j)|yt(i,j)∈mbt,nAnd yt-1(i,j)∈mbt-1,n}
Wherein, cov, std are represented respectively and are sought covariance, seek mean square deviation;yt(i, j) represents the i-th row of current encoded frame jth row Brightness value;yt-1(i, j) represents the brightness value of the i-th row of former frame jth row;mbt,nRepresent present encoding basic block;mbt-1,nTable Show that former frame same position encodes basic block;T represents image time sequence number in the video sequence;N represents basic block in its institute Position number in the picture;ρtt,nRepresent the time correlation degree of present encoding basic block;Basic block refers to the block that encoder is supported Full-size;
Predictive mode judge module, if for judging
(ρtt,n>Const0 and flag=1)Or (ρ tt,n=0 and flag=0), then present encoding basic block is Skip Pattern;
Otherwise, if(ρtt,n>Const1 and flag=1)Or (ρ tt,n<Const3 and flag=0), then currently compile Code basic block is inter-frame forecast mode;
Else if(ρtt,n<Const2 and flag=1)Or (ρ tt,n>Const4 and flag=0), it is current to compile basic Block is intra prediction mode,
Otherwise,(When above-mentioned condition is unsatisfactory for)The current predictive mode for compiling basic block cannot judge.
Wherein, const0, const1, const2, const3, const4 are respectively corresponding decision threshold;It is general desirable 0.99≤const0≤1,0.7<Const1≤1,0≤const2<0.3;Const3=f1 (fps), const4=f2 (fps), f1 For the monotonic decreasing function of frame per second, f2 for frame per second monotonic decreasing function, typically under full frame rate const3≤5, const4 >= 10;Fps represents frame per second.
Further, described first judges that processing meanss include that Skip mode treatment modules, inter-frame forecast mode process mould Block, intra prediction mode processing module, cannot determinating mode processing module, Fig. 7 is the first structure for judging processing meanss in Fig. 5 Figure;
Skip mode treatment modules are right first if being Skip patterns for the quick judged result of the first predictive mode Present encoding basic block carries out Skip predictions, subsequently into next code module;
Inter-frame forecast mode processing module, if being inter-frame forecast mode for the quick judged result of the first predictive mode, Then first inter prediction is carried out to present encoding basic block, subsequently into optimum prediction mode acquisition module;
Intra prediction mode processing module, if being intra prediction mode for the quick judged result of the first predictive mode, Then it is directly entered block complicated dynamic behaviour module;
Cannot determinating mode processing module, if cannot sentence for predictive mode for the quick judged result of the first predictive mode Mould-fixed, then carry out inter prediction to present encoding basic block, subsequently into block complicated dynamic behaviour module.
Further, the first fast intra-mode prediction module includes encoding block division module, fast intra-mode prediction module, Fig. 8 For the structure chart of the first fast intra-mode prediction module in Fig. 5;
Encoding block division module, for present encoding basic block to be divided into the encoding block less than basic block size;
Fast intra-mode prediction module, for adopting in embodiment one to the encoding block each Suo Shu less than basic block size The described Adaptive Mode Selection Method for Intra-Prediction based on correlation analysis carries out infra-frame prediction;
It will be understood by those skilled in the art that realize that all or part of step in above-described embodiment method is can With what is completed by programmed instruction related hardware, described program can be stored in a computer read/write memory medium, Described storage medium can be ROM, RAM, disk, CD etc..
Presently preferred embodiments of the present invention is the foregoing is only, not to limit the present invention, all essences in the present invention Any modification, equivalent and improvement made within god and principle etc., should be included within the scope of the present invention.

Claims (17)

1. a kind of Adaptive Mode Selection Method for Intra-Prediction based on correlation analysis, it is characterised in that described based on correlation The fast intra-frame predicting method of analysis specifically includes following steps:
S1061:Calculate the directional correlation coefficient of present encoding block;
S1062:According to the different situations of the directional correlation coefficient of present encoding block, corresponding intra prediction mode is selected as most Good intra prediction mode Candidate Set;
S1063:The infra-frame prediction of the optimal intra prediction mode Candidate Set correspondence direction is carried out to present encoding block;
In step S1061, the directional correlation coefficient for calculating present encoding block includes:
(A) if the predictive mode support of present encoding block and only all prediction sides of the 4x4 intra prediction modes of support h264 To, then the vertical of present encoding block, level, right diagonal, left diagonally opposed coefficient correlation are calculated, its computational methods is as follows:
Y b D D R ( k 1 ) = Y b ( i , j | i + j = k 1 ) , Y b D D R ( k 2 ) = Y b ( i , j | i + j = k 2 )
Y b D D L ( k 1 ) = Y b ( i , j | j - i = k 1 ) , Y b D D L ( k 2 ) = Y b ( i , j | j - i = k 2 )
Wherein, cov, std, mean (variable | condition) are represented respectively and are sought covariance, ask mean square deviation, the variable to meeting condition to ask Average;YbRepresent the brightness value of present encoding block;Yb(i1,:) represent the i-th of present encoding block1Row brightness value;Yb(i2,:) represent The i-th of present encoding block2Row brightness value;Yb(:,j1) represent present encoding block jth1Row row brightness value;Yb(:,j2) represent and work as The jth of front encoding block2Row row brightness value;Yb(i, j | condition) represent the brightness that the i-th row of the present encoding block jth for meeting condition is arranged Value;Size (b) represents present encoding block row size;rV、rH、rDDR、rDDLRepresent that the vertical of present encoding block, level, the right side are right respectively Angle, left diagonally opposed coefficient correlation, its subscript represents respectively corresponding prediction direction;rV,i1,i2、rH,j1,j2、rDDR,k1,k2、 rDDL,k1,k2, respectively represent vertical, level, right diagonal, left diagonally opposed coefficient correlation intermediate variable;k1、k2Represent respectively First, second ranks and difference variable;
(B) if the predictive mode of present encoding block only supports a part of direction in the 4x4 intra prediction modes of h264, no The corresponding related direction coefficient correlation of predictive mode of support deletes phase strain therein without calculating in corresponding computing formula Amount,
Tetra- kinds of predictive modes of H, V, DC, Plan are only supported such as present encoding block, then now only need to calculate present encoding block it is vertical, Horizontal direction coefficient correlation, its computational methods are as follows:
(C) if the predictive mode of present encoding block is also propped up in addition to prediction direction in the 4x4 intra prediction modes for supporting h264 The prediction direction more segmented is held, then selects following methods one or method two,
Method one:Using process step in step (A);
Method two:The coefficient correlation in directional correlation coefficient and subdivision direction in calculation procedure (A);
Wherein, directional correlation coefficient is done and is analogized by " coefficient correlation in subdivision direction " method according to the method in step (A),
The computational methods of the coefficient correlation in the subdivision direction include:
If the predictive mode of present encoding block is in addition to comprising HD, also comprising the prediction direction between DDR and HD, comprising HD and H Between prediction direction,
,
Wherein
Y b H D ( k 1 ) = Y b ( i , j | i + 2 * j = k 1 ) , Y b H D ( k 2 ) = Y b ( i , j | i + 2 * j = k 2 )
rHDThe coefficient correlation in HD directions is represented, its subscript represents respectively corresponding prediction direction, rHD,k1,k2Represent the phase in HD directions The intermediate variable of relation number;k1、k2First, second ranks and difference variable are represented respectively.
2. the Adaptive Mode Selection Method for Intra-Prediction of correlation analysis is based on as claimed in claim 1, it is characterised in that step It is described " according to the different situations of the directional correlation coefficient of present encoding block, to select corresponding intra prediction mode in rapid S1062 As optimal intra prediction mode Candidate Set " include:
If the predictive mode support of (a) present encoding block and only all prediction sides of the 4x4 intra prediction modes of support h264 To then
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL) corresponding predictive mode, maxless (rV,rH,rDDR,rDDL) corresponding predictive mode,
And corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 pairs between above-mentioned two prediction direction The predictive mode answered }
Situation 3:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL) corresponding predictive mode,
Positioned at above-mentioned prediction direction and max less (rV,rH,rDDR,rDDL) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 4:If
Then optimal intra prediction mode Candidate Set
={ max less (rV,rH,rDDR,rDDL) corresponding predictive mode, positioned at above-mentioned prediction direction with
max(rV,rH,rDDR,rDDL) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL) correspondence prediction direction with
max(rV,rH,rDDR,rDDL) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 6:If above-mentioned situation is all unsatisfactory for, optimal intra prediction mode Candidate Set={ all predictive modes };
Wherein, max (variable) is represented and is sought the maximum for being not equal to 2 variables;Max less (variable) are represented to ask and are not equal to 2 variables Secondary maximum;Abs is represented and is sought absolute value;rinv_maxRepresent perpendicular to max (rV,rH,rDDR,rDDL) correspond to correspondence in prediction direction Coefficient correlation;rinv_maxlessRepresent perpendicular to maxless (rV,rH,rDDR,rDDL) correspond to corresponding phase relation in prediction direction Number;Thres2Represent the second decision threshold, Thres3Represent the 3rd decision threshold, Thres4Represent the 4th decision threshold, Thres5 Represent the 5th decision threshold, Thres2>0.95, Thres4<0.3<Thres5<0.5<Thres3;" directionless predictive mode " refers to nothing Method sorts out the predictive mode of prediction direction;
If b the predictive mode of () present encoding block only supports a part of direction in the 4x4 intra prediction modes of h264, press Relevant variable is removed in " optimal intra prediction mode Candidate Set " calculation procedure according to the method in step (a);
If c the predictive mode of () present encoding block is also propped up in addition to prediction direction in the 4x4 intra prediction modes for supporting h264 The prediction direction more segmented is held, then following methods three or method four may be selected,
Method three:Using process step in step (a);
Method four:Increase in " optimal intra prediction mode Candidate Set " calculation procedure according to the method in step (a) and mutually strain Amount.
3. the Adaptive Mode Selection Method for Intra-Prediction of correlation analysis is based on as claimed in claim 2, it is characterised in that institute Stating step (b) includes:
As present encoding block only supports tetra- kinds of predictive modes of H, V, DC, Plan, now only need to exist according to the method in step (a) Relevant variable is removed in " optimal intra prediction mode Candidate Set " calculation procedure, it is specific as follows:
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH) corresponding predictive mode, max less (rV,rH) corresponding predictive mode,
And corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 pairs between above-mentioned two prediction direction The predictive mode answered }
Situation 3:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH) corresponding predictive mode,
Positioned at above-mentioned prediction direction and max less (rV,rH) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 4:If
Then optimal intra prediction mode Candidate Set
={ max less (rV,rH) corresponding predictive mode, positioned at above-mentioned prediction direction with
max(rV,rH) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL) correspondence prediction direction with
max(rV,rH,rDDR,rDDL) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 6:If above-mentioned situation is all unsatisfactory for, optimal intra prediction mode Candidate Set=All predictive modes};
In this case, rinv_maxRepresent perpendicular to max (rV,rH) correspond to corresponding coefficient correlation in prediction direction; rinv_maxlessRepresent perpendicular to maxless (rV,rH) correspond to corresponding coefficient correlation in prediction direction.
4. the Adaptive Mode Selection Method for Intra-Prediction of correlation analysis is based on as claimed in claim 2, it is characterised in that institute Stating step (c) includes:
If the predictive mode of present encoding block is in addition to comprising HD, also comprising the prediction direction between DDR and HD, comprising HD and H Between prediction direction, now, increase in " optimal intra prediction mode Candidate Set " calculation procedure according to the method in step (a) Plus relevant variable, it is specific as follows
Situation 1:IfThen
Optimal intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, coefficient correlation be equal to 2 corresponding predictive modes
Situation 2:IfIt is then optimal Intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, maxless (rV,rH,rDDR,rDDL,rHD) corresponding prediction Pattern,
And corresponding predictive mode, directionless predictive mode, coefficient correlation are equal to 2 pairs between above-mentioned two prediction direction The predictive mode answered }
Situation 3:IfIt is then optimal Intra prediction mode Candidate Set
={ max (rV,rH,rDDR,rDDL,rHD) corresponding predictive mode,
Positioned at above-mentioned prediction direction and max less (rV,rH,rDDR,rDDL,rHD) correspond to corresponding prediction mould between prediction direction Formula,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 4:If
Then optimal intra prediction mode Candidate Set
={ max less (rV,rH,rDDR,rDDL,rHD) corresponding predictive mode, positioned at above-mentioned prediction direction with
max(rV,rH,rDDR,rDDL,rHD) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 5:If
Then optimal intra prediction mode Candidate Set
={ it is located at max less (rV,rH,rDDR,rDDL,rHD) correspondence prediction direction with
max(rV,rH,rDDR,rDDL,rHD) correspondence prediction direction between corresponding predictive mode,
Directionless predictive mode, coefficient correlation are equal to 2 corresponding predictive modes }
Situation 6:If above-mentioned situation is all unsatisfactory for,
Optimal intra prediction mode Candidate Set={ all predictive modes },
In this case, rinv_maxRepresent perpendicular to max (rV,rH,rDDR,rDDL,rHD) correspond to corresponding correlation in prediction direction Coefficient;rinv_maxlessRepresent perpendicular to maxless (rV,rH,rDDR,rDDL,rHD) correspond to corresponding coefficient correlation in prediction direction.
5. a kind of predictive mode fast selecting method based on correlation analysis, it is characterised in that methods described includes following step Suddenly,
S101:Judge whether present frame is inter-frame, if yes then enter S102;
S102:The first predictive mode is carried out to present encoding basic block quickly to judge;
S103c:If the quick judged result of the first predictive mode is intra prediction mode, step S104 is directly entered;
S104:Calculate the block complexity of present encoding basic block;
S105:Judge whether mbit,n>Thres1, S106 is if so, then entered, if it is not, S107 is then entered,
Wherein, mbit,nThe block complexity of present encoding basic block is represented, t represents image time sequence number in the video sequence;n Represent basic block its position number in the picture;Thres1Represent the first decision threshold;
S106:Present encoding basic block is divided into into the encoding block less than basic block size, to the encoding block using such as right It is required that the Adaptive Mode Selection Method for Intra-Prediction based on correlation analysis described in 1 carries out infra-frame prediction;
S107:Using the Adaptive Mode Selection Method for Intra-Prediction based on correlation analysis as claimed in claim 1 to current Coding basic block carries out infra-frame prediction;
S108:RDO is carried out to present encoding basic block, optimum prediction mode is obtained;
S109:Change quantization, scanning are carried out to present encoding basic block, entropy code operation completes the volume of present encoding basic block Code;
S110:Judge that next coding basic block whether there is, if it is present next coding basic block is set to into present encoding Block, reenters S102;Otherwise judge that next coded frame whether there is, if it is present next coded frame is set to currently Coded frame, reenters S101;Otherwise, terminate.
6. the predictive mode fast selecting method of correlation analysis is based on as claimed in claim 5, it is characterised in that the step It is further comprising the steps of after rapid S102,
S103d:If the quick judged result of the first predictive mode be predictive mode cannot determinating mode, it is basic to present encoding Block carries out inter prediction, subsequently into step S104.
7. the predictive mode fast selecting method of correlation analysis is based on as claimed in claim 5, it is characterised in that the step It is further comprising the steps of after rapid S102,
S103a:If the quick judged result of the first predictive mode is Skip patterns, present encoding basic block is carried out first Skip is predicted, subsequently into S109.
8. the predictive mode fast selecting method of correlation analysis is based on as claimed in claim 5, it is characterised in that the step It is further comprising the steps of after rapid S102,
S103b:If the quick judged result of the first predictive mode is inter-frame forecast mode, present encoding basic block is entered first Row inter prediction, subsequently into S108.
9. the predictive mode fast selecting method of correlation analysis is based on as claimed in claim 5, it is characterised in that the step In rapid S101,
Judge whether present frame is inter-frame, if it is not, then into step S104.
10. the predictive mode fast selecting method of correlation analysis is based on as claimed in claim 5, it is characterised in that first Predictive mode quick judgment method includes:
S1020:The symbolic variable flag of present encoding basic block is calculated,
S1021:The time correlation degree of present encoding basic block is calculated,
Yt,n={ yt(i,j)|yt(i,j)∈mbt,n}、Yt-1,n={ yt-1(i,j)|yt-1(i,j)∈mbt-1,n}
dYt,n={ yt(i,j)-yt-1(i,j)|yt(i,j)∈mbt,nAnd yt-1(i,j)∈mbt-1,n}
Wherein, cov, std are represented respectively and are sought covariance, seek mean square deviation;yt(i, j) represents the bright of the i-th row of current encoded frame jth row Angle value;yt-1(i, j) represents the brightness value of the i-th row of former frame jth row;mbt,nRepresent present encoding basic block;mbt-1,nBefore expression One frame same position encodes basic block;T represents image time sequence number in the video sequence;N represents basic block in its place figure Position number as in;ρtt,nRepresent the time correlation degree of present encoding basic block;Basic block refers to the block of encoder support most Large scale;S1022:If (ρ tt,n>Const0 and flag=1) or (ρ tt,n=0 and flag=0), then present encoding Basic block is Skip patterns;
Otherwise, if (ρ tt,n>Const1 and flag=1) or (ρ tt,n<Const3 and flag=0), then present encoding Basic block is inter-frame forecast mode;
Else if (ρ tt,n<Const2 and flag=1) or (ρ tt,n>Const4 and flag=0), currently compile basic block For intra prediction mode;
Otherwise, when above-mentioned condition is unsatisfactory for, the current predictive mode for compiling basic block cannot judge;
Wherein, const 0, const 1, const 2, const 3, const 4 respectively corresponding decision threshold, 0.99≤ Const0≤1,0.7<Const1≤1,0≤const2<0.3;Const3=f1 (fps), const4=f2 (fps), f1 is frame The monotonic decreasing function of rate, f2 for frame per second monotonic decreasing function, const3≤5, const 4 >=10 under full frame rate;Fps tables Show frame per second.
The 11. predictive mode fast selecting methods based on correlation analysis as claimed in claim 5, it is characterised in that described The block complexity for calculating present encoding basic block is comprised the following steps:
S1041:The boundary intensity of each pixel in present encoding basic block is calculated,
yit(i, j)=| | filter (yt(i,j)|yt(i,j)∈mbt,n)||;
S1042:Mean square deviation is asked to the boundary intensity of each pixel in present encoding basic block
mbit,n=std (yit(i,j)|yt(i,j)∈mbt,n),
Wherein, yit(i, j) represents pixel ytThe boundary intensity of (i, j);yt(i, j) represents the bright of the i-th row of current encoded frame jth row Angle value;Std is represented seek mean square deviation respectively;mbt,nRepresent present encoding basic block;Filter (variable | condition) represent to meeting bar The variable of part carries out boundary filtering;| | variable | | is represented to variable modulus.
A kind of 12. intra prediction modes based on correlation analysis quickly select system, it is characterised in that the system includes side Module, optimal intra-framed prediction module are selected to coefficient correlation computing module, optimal intra prediction mode candidate,
Directional correlation coefficients calculation block, for calculating the directional correlation coefficient of present encoding block;
Optimal intra prediction mode candidate selects module, for according to the different situations of the directional correlation coefficient of present encoding block, choosing Corresponding intra prediction mode is selected as optimal intra prediction mode Candidate Set;
Optimal intra-framed prediction module, for carrying out the frame in of optimal intra prediction mode Candidate Set correspondence direction to present encoding block Prediction;
The directional correlation coefficient for calculating present encoding block includes:
(A) if the predictive mode support of present encoding block and only all prediction sides of the 4x4 intra prediction modes of support h264 To, then the vertical of present encoding block, level, right diagonal, left diagonally opposed coefficient correlation are calculated, its computational methods is as follows:
Y b D D R ( k 1 ) = Y b ( i , j | i + j = k 1 ) , Y b D D R ( k 2 ) = Y b ( i , j | i + j = k 2 )
Y b D D L ( k 1 ) = Y b ( i , j | j - i = k 1 ) , Y b D D L ( k 2 ) = Y b ( i , j | j - i = k 2 )
Wherein, cov, std, mean (variable | condition) are represented respectively and are sought covariance, ask mean square deviation, the variable to meeting condition to ask Average;YbRepresent the brightness value of present encoding block;Yb(i1,:) represent the i-th of present encoding block1Row brightness value;Yb(i2,:) represent The i-th of present encoding block2Row brightness value;Yb(:,j1) represent present encoding block jth1Row row brightness value;Yb(:,j2) represent and work as The jth of front encoding block2Row row brightness value;Yb(i, j | condition) represent the brightness that the i-th row of the present encoding block jth for meeting condition is arranged Value;Size (b) represents present encoding block row size;rV、rH、rDDR、rDDLRepresent that the vertical of present encoding block, level, the right side are right respectively Angle, left diagonally opposed coefficient correlation, its subscript represents respectively corresponding prediction direction;rV,i1,i2、rH,j1,j2、rDDR,k1,k2、 rDDL,k1,k2, respectively represent vertical, level, right diagonal, left diagonally opposed coefficient correlation intermediate variable;k1、k2Represent respectively First, second ranks and difference variable;
(B) if the predictive mode of present encoding block only supports a part of direction in the 4x4 intra prediction modes of h264, no The corresponding related direction coefficient correlation of predictive mode of support deletes phase strain therein without calculating in corresponding computing formula Amount,
Tetra- kinds of predictive modes of H, V, DC, Plan are only supported such as present encoding block, then now only need to calculate present encoding block it is vertical, Horizontal direction coefficient correlation, its computational methods are as follows:
(C) if the predictive mode of present encoding block is also propped up in addition to prediction direction in the 4x4 intra prediction modes for supporting h264 The prediction direction more segmented is held, then selects following methods one or method two,
Method one:Using process step in step (A);
Method two:The coefficient correlation in directional correlation coefficient and subdivision direction in calculation procedure (A);
Wherein, directional correlation coefficient is done and is analogized by " coefficient correlation in subdivision direction " method according to the method in step (A),
The computational methods of the coefficient correlation in the subdivision direction include:
If the predictive mode of present encoding block is in addition to comprising HD, also comprising the prediction direction between DDR and HD, comprising HD and H Between prediction direction,
,
Wherein
Y b H D ( k 1 ) = Y b ( i , j | i + 2 * j = k 1 ) , Y b H D ( k 2 ) = Y b ( i , j | i + 2 * j = k 2 )
rHDThe coefficient correlation in HD directions is represented, its subscript represents respectively corresponding prediction direction, rHD,k1,k2Represent the phase in HD directions The intermediate variable of relation number;k1、k2First, second ranks and difference variable are represented respectively.
A kind of 13. predictive modes based on correlation analysis quickly select system, it is characterised in that the system includes:
Inter-frame judge module, for judging whether present frame is inter-frame, quickly sentences if yes then enter the first predictive mode Disconnected module, if otherwise entering block complicated dynamic behaviour module;
The quick judge module of first predictive mode, quickly judges for carrying out the first predictive mode to present encoding basic block;
First judges processing meanss, for according to the quick judged result of the first predictive mode, at present encoding basic block Reason;
Block complicated dynamic behaviour module, for calculating the block complexity of present encoding basic block;
5th judge module, for judging whether mbit,n>Thres1, if so, then enter the first fast intra-mode prediction module;If It is no, then into the second fast intra-mode prediction module;
mbit,nRepresent the block complexity of present encoding basic block, Thres1The first decision threshold is represented,
Thres1>25;T represents image time sequence number in the video sequence;N represent basic block its position in the picture Sequence number;
First fast intra-mode prediction module, for present encoding basic block to be divided into the encoding block less than basic block size, so Each encoding block is carried out using the intra prediction mode fast method based on correlation analysis as claimed in claim 1 afterwards Infra-frame prediction;
Second fast intra-mode prediction module, for pre- using the frame in based on correlation analysis as claimed in claim 1 Survey pattern fast method carries out infra-frame prediction to present encoding basic block;
Optimum prediction mode acquisition module, judges that processing meanss and next code module are connected, for present encoding with first Basic block carries out RDO, obtains optimum prediction mode;
Next code module, judges that processing meanss and next coding basic block judging treatmenting module are connected with first, for working as Front coding basic block carries out change quantization, scanning, and entropy code operation completes the coding of present encoding basic block;
Next coding basic block judging treatmenting module, for judging that next coding basic block whether there is, if it is present by under One coding basic block is set to present encoding block, reenters the quick judge module of the first predictive mode;Otherwise, into next volume Code frame judging treatmenting module;
Next coded frame judging treatmenting module, for judging that next coded frame whether there is, if it is present by next coded frame Current encoded frame is set to, inter-frame judge module is reentered;Otherwise, terminate.
14. predictive modes based on correlation analysis as claimed in claim 13 quickly select system, it is characterised in that block is multiple Miscellaneous degree computing module, including
Boundary strength calculation module, for calculating the boundary intensity of each pixel in present encoding basic block,
yit(i, j)=| | filter (yt(i,j)|yt(i,j)∈mbt,n)||
Boundary intensity mean square deviation computing module, for seeking mean square deviation to the boundary intensity of each pixel in present encoding basic block,
mbit,n=std (yit(i,j)|yt(i,j)∈mbt,n)
Wherein, yit(i, j) represents pixel ytThe boundary intensity of (i, j);yt(i, j) represents the bright of the i-th row of current encoded frame jth row Angle value;Std is represented seek mean square deviation respectively;mbt,nRepresent present encoding basic block;Filter (variable | condition) represent to meeting bar The variable of part carries out boundary filtering;| | variable | | is represented to variable modulus;mbit,nRepresent that the block of present encoding basic block is complicated Degree.
15. predictive modes based on correlation analysis as claimed in claim 13 quickly select system, it is characterised in that described The quick judge module of first predictive mode also includes that symbolic variable computing module, time correlation degree computing module, predictive mode are sentenced Disconnected module,
Symbolic variable computing module, for calculating the symbolic variable flag of present encoding basic block;
Time correlation degree computing module, for calculating the time correlation degree of present encoding basic block;
Yt,n={ yt(i,j)|yt(i,j)∈mbt,n}、Yt-1,n={ yt-1(i,j)|yt-1(i,j)∈mbt-1,n}
dYt,n={ yt(i,j)-yt-1(i,j)|yt(i,j)∈mbt,nAnd yt-1(i,j)∈mbt-1,n}
Wherein, cov, std are represented respectively and are sought covariance, seek mean square deviation;yt(i, j) represents the bright of the i-th row of current encoded frame jth row Angle value;yt-1(i, j) represents the brightness value of the i-th row of former frame jth row;mbt,nRepresent present encoding basic block;mbt-1,nBefore expression One frame same position encodes basic block;T represents image time sequence number in the video sequence;N represents basic block in its place figure Position number as in;ρtt,nRepresent the time correlation degree of present encoding basic block;Basic block refers to the block of encoder support most Large scale;
Predictive mode judge module, if for judging
(ρtt,n>Const0 and flag=1) or (ρ tt,n=0 and flag=0), then present encoding basic block is Skip moulds Formula;
Otherwise, if (ρ tt,n>Const1 and flag=1) or (ρ tt,n<Const3 and flag=0), then present encoding Basic block is inter-frame forecast mode;
Else if (ρ tt,n<Const2 and flag=1) or (ρ tt,n>Const4 and flag=0), currently compile basic block For intra prediction mode,
Otherwise, when above-mentioned condition is unsatisfactory for, the current predictive mode for compiling basic block cannot judge,
Wherein, const0, const1, const2, const3, const4 are respectively corresponding decision threshold;
0.99≤const0≤1,0.7<Const1≤1,0≤const2<0.3;Const3=f1 (fps), const4=f2 (fps), f1 for frame per second monotonic decreasing function, f2 for frame per second monotonic decreasing function, under full frame rate const3≤5, const4≥10;Fps represents frame per second.
16. predictive modes based on correlation analysis as claimed in claim 13 quickly select system, it is characterised in that
Described first judges that processing meanss include Skip mode treatment modules, inter-frame forecast mode processing module, infra-frame prediction mould Formula processing module, cannot determinating mode processing module,
Skip mode treatment modules, if being Skip patterns for the quick judged result of the first predictive mode, first to current Coding basic block carries out Skip predictions, subsequently into next code module;
Inter-frame forecast mode processing module, it is first if being inter-frame forecast mode for the quick judged result of the first predictive mode First inter prediction is carried out to present encoding basic block, subsequently into optimum prediction mode acquisition module;
Intra prediction mode processing module, if being intra prediction mode for the quick judged result of the first predictive mode, directly Tap into into block complicated dynamic behaviour module;
Cannot determinating mode processing module, if mould cannot be judged for predictive mode for the quick judged result of the first predictive mode Formula, then carry out inter prediction to present encoding basic block, subsequently into block complicated dynamic behaviour module.
17. predictive modes based on correlation analysis as claimed in claim 13 quickly select system, it is characterised in that
First fast intra-mode prediction module includes encoding block division module, fast intra-mode prediction module,
Encoding block division module, for present encoding basic block to be divided into the encoding block less than basic block size;
Fast intra-mode prediction module, for adopting the encoding block each Suo Shu less than basic block size such as claim 1 institute That what is stated carries out infra-frame prediction based on the intra prediction mode fast method of correlation analysis.
CN201310574825.XA 2013-11-18 2013-11-18 Fast prediction mode selection method and system based on correlation analysis Expired - Fee Related CN103686166B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310574825.XA CN103686166B (en) 2013-11-18 2013-11-18 Fast prediction mode selection method and system based on correlation analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310574825.XA CN103686166B (en) 2013-11-18 2013-11-18 Fast prediction mode selection method and system based on correlation analysis

Publications (2)

Publication Number Publication Date
CN103686166A CN103686166A (en) 2014-03-26
CN103686166B true CN103686166B (en) 2017-05-17

Family

ID=50322256

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310574825.XA Expired - Fee Related CN103686166B (en) 2013-11-18 2013-11-18 Fast prediction mode selection method and system based on correlation analysis

Country Status (1)

Country Link
CN (1) CN103686166B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104363450B (en) * 2014-11-27 2017-10-27 北京奇艺世纪科技有限公司 A kind of intra-frame encoding mode decision-making technique and device
CN108259913A (en) 2016-12-29 2018-07-06 北京大学深圳研究生院 A kind of intra-frame prediction method in MB of prediction frame
CN110545402A (en) * 2019-08-18 2019-12-06 宁波职业技术学院 A downhole monitoring video processing method, computer equipment and storage medium
CN116708934B (en) * 2023-05-16 2024-03-22 深圳东方凤鸣科技有限公司 Video coding processing method and device

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101309408A (en) * 2008-06-13 2008-11-19 青岛海信电子产业控股股份有限公司 Lightness block selection method of intra-frame prediction mode
CN101895761A (en) * 2010-07-29 2010-11-24 江苏大学 Quick intraframe prediction algorithm
CN102387364A (en) * 2011-10-27 2012-03-21 江苏大学 Fast intra-frame mode selecting algorithm

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2013179572A (en) * 2012-02-08 2013-09-09 Canon Inc Image encoding apparatus and image encoding method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101309408A (en) * 2008-06-13 2008-11-19 青岛海信电子产业控股股份有限公司 Lightness block selection method of intra-frame prediction mode
CN101895761A (en) * 2010-07-29 2010-11-24 江苏大学 Quick intraframe prediction algorithm
CN102387364A (en) * 2011-10-27 2012-03-21 江苏大学 Fast intra-frame mode selecting algorithm

Also Published As

Publication number Publication date
CN103686166A (en) 2014-03-26

Similar Documents

Publication Publication Date Title
Saldanha et al. Configurable fast block partitioning for VVC intra coding using light gradient boosting machine
CN106961606B (en) HEVC intra-frame coding mode selection method based on texture division characteristics
CN104320666B (en) Image processing equipment and method
CN109905712A (en) Fast Mode Decision Algorithm for HEVC Intra-coding Based on ResNet
CN103188496B (en) Based on the method for coding quick movement estimation video of motion vector distribution prediction
CN100594732C (en) An inter-frame mode selection method
CN101583036B (en) A method for determining the relationship between motion features and high-efficiency coding modes in pixel-domain video transcoding
CN101325707B (en) Texture Adaptive Video Codec System
CN103686166B (en) Fast prediction mode selection method and system based on correlation analysis
CN102090065A (en) Image encoding device, image decoding device, image encoding method, and image decoding method
CN100574447C (en) Fast intraframe predicting mode selecting method based on the AVS video coding
CN101175212A (en) Low-complexity in-frame estimation mode selection method
CN102714732A (en) Encoding device, decoding device and data structure
CN111988628B (en) Fast intra-frame coding method based on reinforcement learning VVC
US20110129013A1 (en) Method and apparatus for adaptively determining compression modes to compress frames
CN101304529A (en) Method and device for selecting macro block mode
CN103533355A (en) Quick coding method for HEVC (high efficiency video coding)
CN100596202C (en) Fast mode selection method in frame
CN101588487B (en) A video intra-frame predictive coding method
CN116634137A (en) Fast decision-making of screen content video intra-mode based on feature intersection
CN103533349A (en) Support vector machine-based fast inter-frame prediction macro block mode selection method for B frame
CN101867818B (en) Selection method and device of macroblock mode
US20080253670A1 (en) Image Signal Re-Encoding Apparatus And Image Signal Re-Encoding Method
US8175150B1 (en) Methods and/or apparatus for implementing rate distortion optimization in video compression
CN103517077A (en) Method and device for rapidly selecting prediction mode

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C53 Correction of patent for invention or patent application
CB02 Change of applicant information

Address after: The central Shenzhen city of Guangdong Province, 518057 Keyuan Road, Nanshan District science and Technology Park No. 15 Science Park Sinovac A Building 1 unit 403, No. 405 unit

Applicant after: Shenzhen Yunzhou Multimedia Technology Co., Ltd.

Address before: Unit B4 9 building 518057 Guangdong city of Shenzhen province Nanshan District high in the four EVOC Technology Building No. 31

Applicant before: Shenzhen Yunzhou Multimedia Technology Co., Ltd.

GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20170517

Termination date: 20191118

CF01 Termination of patent right due to non-payment of annual fee