CN101188755B - A method for VBR code rate control in AVX decoding of real time video signals - Google Patents
A method for VBR code rate control in AVX decoding of real time video signals Download PDFInfo
- Publication number
- CN101188755B CN101188755B CN 200710161158 CN200710161158A CN101188755B CN 101188755 B CN101188755 B CN 101188755B CN 200710161158 CN200710161158 CN 200710161158 CN 200710161158 A CN200710161158 A CN 200710161158A CN 101188755 B CN101188755 B CN 101188755B
- Authority
- CN
- China
- Prior art keywords
- image
- group
- current frame
- image group
- actual
- 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
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a VBR bit rate control method of a real-time video signal in the AVS coding process. According to the forecasting complexity of the current image group, the object bit rate of the current image group is allocated, object bit numbers are equally distributed to an image in the current image group, according to a rate distortion model, and the quantification parameter of the image is calculated. When the scene change is detected, the corresponding residual bit numbers in the image group where the image with abrupt complexity is positioned are updated. the accuracy of the bit rate control is ensured, the defined bit rate can be better approached, the video quality is improved, the overall quality reduction caused by the quality fluctuation is avoided, thus ensuring thatthe smoothing property of the video quality is better; the invention can better distribute the object bit rate of the current image group according to the forecasting complexity of the current image group, and has stronger adaptability to message source; the invention has the advantages that the pre-coding analysis is not required to the imaged, and the coding is operated once, and has the advantages that the calculation complexity is low, the calculation intensity is low, the decoding complexity can be decreased, and the real time is improved.
Description
Technical field
The present invention relates to a kind of video signal processing method, especially relate to a kind of real time video signals VBR code rate control method in the AVS cataloged procedure.
Background technology
AVS (Audio Video coding Standard, digital audio/video encoding and decoding technique standard) be that Chinese first has the second generation digital audio/video encoding and decoding standard of independent intellectual property right, full name is " an information technology advanced audio/video encoding standard ", its code efficiency and competitive suitable with international standard MPEG-4/H.264 has been represented current international most advanced level.
Rate Control is one of important technology of video coding, be in video transmission, video storage and design process of hardware, all play a part very important.It is source properties according to video sequence that VBR (Variable Bit Rate, variable bit rate) Rate Control is coded under the perfect condition, allocated code bit number dynamically, and it is constant to make that coding produces video quality.Each international video encoding and decoding standard all has its typical bit rate control method separately, as: the VM8 of the TM5 of MPEG-2, TMN8 H.263, MPEG-4 and JVT-H017 H.264, because AVS and H.264 very big similitude is arranged, so the bit rate control method that AVS takes in identifying code rm52j also adopts JVT-H017.
The main process of JVT-H017 bit rate control method is as follows:
(1) according to the target bit of fluid flow blockage Model Calculation current frame image;
(2) be the remaining bit number of uncoded elementary cell mean allocation in the current frame image;
(3) according to the MAD of uncoded elementary cell in the MAD of same position elementary cell in MAD (Mean Absolute Difference, the quadratic sum of difference) linear prediction model and the former frame image prediction current frame image;
(4) calculate the quantization parameter of uncoded elementary cell in the current frame image according to the secondary rate-distortion model;
(5) according to the quantization parameter that calculates each MB in the elementary cells all in the current frame image (Macro-block, macro block) is done RDO (Rate distortion optimization, rate-distortion optimization) computing.
This method has adopted based on the secondary rate-distortion model of Taylor series expansion and MAD linear prediction model, in implementation procedure,, the sampling point in the window of variable-size fits by being carried out function, upgrade secondary R-D (Rate Distortion, rate distortion) model and MAD linear prediction model parameter, then remove the bad point in the window, upgrade secondary R-D model and MAD linear prediction model parameter once more.Wherein, function fits general employing least square method, carry out matrix multiple computing repeatedly, and the size of matrix is by the size decision of window.This method makes Rate Control more accurate, and can be good at approaching the target bit rate of qualification owing to introduced secondary R-D model and MAD linear prediction model; In addition, utilize this method can produce mild relatively video quality.But there is significant disadvantages simultaneously in this method:
1) introducing of secondary R-D model and MAD linear prediction model has increased computation complexity simultaneously, for flush bonding processor, the loop computation that multiplication occupies will be used real-time video coding much larger than general plus-minus, shift operation, is a very big challenge to flush bonding processor;
2) operand of video coding own is just than higher, add the high operand that adopts in the calculating process based on macro block, in conjunction with the bit rate control method of RDO, this will become the bottleneck that encoder is used in real time;
3) because the distribution method and the hypothetical reference decoder buffering of the code check of this algorithm GOP (Group of Picture, image sets) have limited the further raising of the mild degree of front and back quality.
Summary of the invention
Technical problem to be solved by this invention provide a kind of can guarantee that the video quality that produces is mild more and simple, computation complexity is low to real time video signals VBR code rate control method in the AVS cataloged procedure.
The present invention solves the problems of the technologies described above the technical scheme that is adopted: a kind of to real time video signals VBR code rate control method in the AVS cataloged procedure, signal pickup assembly obtains vision signal and described vision signal is carried out the VBR Rate Control the AVS cataloged procedure from actual scene, described vision signal comprises a plurality of image sets, each image sets comprises an I two field picture and at least one P two field picture, and real time video signals VBR code rate control method in the AVS cataloged procedure be may further comprise the steps:
A. the initial quantization parameter of initialization vision signal is set the initial target code check, and obtain the frame per second of encoder, an image sets comprises the frame number of image and the Gao Yukuan of image, the product of the height and width of note image is the size of image;
B. judge whether the residing present image group of current frame image is first image sets of vision signal, if, then the target bit rate of present image group is the initial target code check, the target bit of present image group is that target bit rate and image sets of present image group comprises the frame number of image and the product of the ratio of frame per second, and execution in step f; Otherwise, continue to carry out;
C. according to the average actual complex degree of previous image group and all the average actual complex degree of image encoded group, through type
Calculate the prediction complexity XGopCurEst of present image group, wherein, XGopPre is the average actual complex degree of previous image group, and XGopAvg is all average actual complex degree of image encoded group;
D. go on foot the prediction complexity XGopCurEst of the present image group that calculates according to c, use formula
Calculate the target bit rate GopBitRate of present image group, wherein, XGopCurEst is the prediction complexity of present image group, PrePQpAVG is the average quantisation parameter of previous image group, XGopAvgCurEst is all average complexities of obtaining of the weighted calculation of the prediction complexity of the actual average complexity of image encoded group and present image group, TotalAvgQp is all average quantisation parameter of image encoded group, and avg_bit_rate is all average actual bit rates of image encoded group;
E. according to the target bit rate GopBitRate of the present image group that calculates in the steps d, calculate the target bit of present image group, its value comprises the frame number of image and the product of the ratio of frame per second for target bit rate and an image sets of present image group;
F. judge whether current frame image is I two field picture or first P two field picture of present image group, if, then when the present image group is first image sets, the quantization parameter of I two field picture or first P two field picture is the initial quantization parameter, and by the initial quantization parameter I two field picture or first P two field picture are encoded execution in step i then; When the present image group is not first image sets, the quantization parameter of I two field picture or first P two field picture is the average quantisation parameter of previous image group, and by the average quantisation parameter of previous image group I two field picture or first P two field picture are encoded execution in step i then; Otherwise, continue to carry out;
G. according to the target bit of present image group, through type
Calculate the target bit FrameBits of current frame image, wherein, RemainBits is the remaining bits number of present image group, its value is that the target bit of present image group and the actual bit that the present image group takies are counted the poor of summation, picnum_Gop is the frame number that an image sets comprises image, and NumberofCodedPFrame is the frame number of all P two field pictures of having encoded of present image group;
H. according to the target bit of the current frame image that calculates in the step g, first percent of pass distortion model
Calculate model parameter C, wherein R is the actual bit number that last P two field picture takies, and X is the actual complex degree of last P two field picture, and QP is the quantization parameter of last P two field picture; Pass through again
Calculate the quantization parameter QP ' of current frame image, wherein X ' is the prediction complexity of current frame image, and its value is the actual complex degree of former frame image, and R ' is the target bit of current frame image; By the quantization parameter of current frame image current frame image is encoded then;
If the i.i-1. intact image of present encoding is the I two field picture, then add up the actual bit number that current frame image takies, the remaining bits number of present image group is the poor of the target bit of present image group and actual bit number that current frame image takies, and the initial value that the actual bit that the present image group takies is counted summation is the actual bit number that current frame image takies; If the i-2. intact image of present encoding is the P two field picture, then add up the actual bit number that current frame image takies, calculate the actual complex degree of current frame image, the actual bit number that its value takies for current frame image and the squared difference of corresponding its reference picture of current frame image and product divided by the size of image, when current frame image is first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the initial value of the actual complex degree summation of present image group is the actual complex degree of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted the actual bit that summation and current frame image take and counted sum that the actual bit that the present image group takies is counted summation, and the initial value of the quantization parameter summation of present image group is the quantization parameter of current frame image; When current frame image is not first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the actual complex degree summation of present image group is the actual complex degree summation of the present image group before the current frame image coding and the actual complex degree sum of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted summation and counted sum with the actual bit that current frame image takies that the actual bit that the present image group takies is counted summation, and the quantization parameter summation of present image group is the quantization parameter summation of the preceding present image group of current frame image coding and the quantization parameter sum of current frame image;
If j. the present image group is not that first image sets and current frame image are the P two field pictures, all the average actual bit number that takies of the every two field picture of image encoded group be all the average actual bit rate of image encoded group and the merchant of frame per second, and whether the product of quantization parameter of judging actual bit number that current frame image takies and current frame image is greater than all 2～3 times of product of the average quantisation parameter of image encoded group or less than all the average actual bit number of image encoded group and all 0.4～0.5 times of product of the average quantisation parameter of image encoded group of the average actual bit number that takies of the every frame of image encoded group and all, if, then determine to detect scene and switch through type
Upgrade the remaining bits number of present image group, RemainBits is the remaining bits number of present image group in the formula, FrameReallyBits is the actual bit number that current frame image takies, and avg_bit_rate is all average actual bit rates of image encoded group, and FRAMERATE is a frame per second;
K. judge whether to exist follow-up vision signal, if exist, judge again then whether the image behind the coding is last image of present image group, if, then actual complex degree summation, the quantization parameter summation of the present image group that obtains according to step I and the actual bit that takies are counted summation, calculate the average quantisation parameter of present image group and average actual complex degree, all average actual complex degree, average quantisation parameter and average actual bit rates of image encoded group, jump to step b then, otherwise, execution in step f; If there is no, then finish.
Compared with prior art, the invention has the advantages that the target bit rate that distributes the present image group by the prediction complexity dynamic self-adapting of present image group, image in the present image group is taked the mean allocation target bit, again according to the quantization parameter of simple rate-distortion model computed image, when image actual complex degree is undergone mutation, the residing image sets of this image chased after subtract or append corresponding remaining bits number.It has not only guaranteed the accuracy of Rate Control, and can approach the code check of qualification preferably, improves video quality, and the oeverall quality of having avoided causing because of quality fluctuation descends, and makes that the mild property of video is better; The present invention can distribute the target bit rate of present image group preferably according to the prediction complexity of present image group, information source is had stronger adaptability; The present invention has the once characteristics of coding, does not need the image precoding is analyzed, and can reduce the complexity of coding, is convenient to hardware and realizes, improves real-time; In addition, computation complexity of the present invention is low, and operand is little, can be widely used in the high fields such as video monitoring that require in real time, preferably resolves contradiction between video quality, real-time and the code check.
Description of drawings
Fig. 1 is the flow chart of the inventive method;
Fig. 2 is the structural representation of vision signal;
Fig. 3 is for adopting the statistical comparison figure of the inventive method and the video quality that adopts the JVT-H017 method to produce during for 1.5Mbps at the initial target code check.
Embodiment
Embodiment describes in further detail the present invention below in conjunction with accompanying drawing.
As depicted in figs. 1 and 2, a kind of to real time video signals VBR code rate control method in the AVS cataloged procedure, signal pickup assembly obtains vision signal and vision signal is carried out the VBR Rate Control in the AVS cataloged procedure encoder from actual scene, vision signal comprises a plurality of image sets, each image sets behind the AVS coding comprises an I two field picture and at least one P two field picture, does not comprise the B two field picture.
The inventive method distributes more bit number for the higher image of complexity, distribute less bit number for the lower image of complexity, according to all the actual complex degree of image encoded group predict the prediction complexity (except that first image sets) of present image group, in conjunction with all actual average code checks of image encoded group, according to the target bit rate of the distribution present image group of present image group prediction complexity dynamic self-adapting; Calculate the target bit of present image group again according to the target bit rate of present image group; According to the target bit of present image group, effectively give each their target bit of image averaging distribution in the image sets then; Calculate the quantization parameter of present image then according to a simple rate distortion R-D model, after the present image of having encoded, calculate the actual bit number that present image takies, scene change detection again, if determining to detect scene switches, then upgrade the remaining bits number of present image group, so that better adapt to true information source feature.The inventive method flow process as shown in Figure 1, it comprises following concrete steps:
A. the initial quantization parameter of initialization vision signal is set the initial target code check, and obtain the frame per second of encoder, an image sets comprises the frame number of image and the Gao Yukuan of image, the product of the height and width of note image is the size of image;
B. judge whether the residing present image group of current frame image is first image sets of vision signal, if, then the target bit rate of present image group is the initial target code check, the target bit of present image group is that target bit rate (initial target code check just) and image sets of present image group comprises the frame number of image and the product of the ratio of frame per second, and execution in step f; Otherwise, continue to carry out;
C. according to the average actual complex degree of previous image group and all the average actual complex degree of image encoded group, through type
Calculate the prediction complexity XGopCurEst of present image group, in conjunction with Fig. 2, in the formula, XGopPre is the average actual complex degree of previous image group (being the image sets that is numbered N-1 among Fig. 2), previous image group and present image group correlation are the strongest, and XGopAvg is all average actual complex degree of image encoded group (being to number all average actual complex degree of coded image group of 0～N-1 among Fig. 2);
D. go on foot the prediction complexity XGopCurEst of the present image group that calculates according to c, through type
Calculate the target bit rate GopBitRate of present image group, in conjunction with Fig. 2, in the formula, XGopCurEst is the prediction complexity of present image group, PrePQpAVG is the average quantisation parameter of previous image group (being the image sets that is numbered N-1 among Fig. 2), TotalAvgQp is all average quantisation parameter of image encoded group (being to number all image sets of 0～N-1 among Fig. 2), avg_bit_rate is all average actual bit rates of image encoded group (being to number all image sets of 0～N-1 among Fig. 2), XGopAvgCurEst is all average complexities of obtaining of the weighted calculation of the prediction complexity of the actual average complexity of image encoded group (being to number all image sets of 0～N-1 among Fig. 2) and present image group, the concrete computational methods of XGopAvgCurEst are XGopAvgCurEst=(XGopAVG* (NumberofGOP-1)+XGopCurEst)/NumberofGOP, XGopAVG is for numbering the average actual complex degree of 0～N-1 all images group among Fig. 2, and NumberofGOP comprises the present image group and the sum of coded image group;
In order to guarantee the convergence of vision signal average bit rate, need be limited to the target bit rate of present image group and to begin initial target code check that the user sets most up and down in the fluctuation range, fluctuating range be generally setting the initial target code check ± 20%, that is: GopBitRate=MIN (MAX (GopBitRate, bit_rate_down), bit_rate_up), wherein the value of bit_rate_down is an initial target code check * (1-20%), and the value of bit_rate_up is an initial target code check * (1+20%);
E. according to the target bit rate GopBitRate of the present image group that calculates in the steps d, calculate the target bit of present image group, its value comprises the frame number of image and the product of the ratio of frame per second for target bit rate and an image sets of present image group;
F. judge whether current frame image is I two field picture or first P two field picture of present image group, if, then when the present image group is first image sets, the quantization parameter of I two field picture or first P two field picture is the initial quantization parameter, and by the initial quantization parameter I two field picture or first P two field picture are encoded execution in step i then; When the present image group is not first image sets, the quantization parameter of I two field picture or first P two field picture is the average quantisation parameter of previous image group, and by the average quantisation parameter of previous image group I two field picture or first P two field picture are encoded execution in step i then; Otherwise, continue to carry out;
G. according to the target bit of present image group, through type
Calculate the target bit FrameBits of current frame image, wherein, RemainBits is the remaining bits number of present image group, its value is that the target bit of present image group and the actual bit that the present image group takies are counted the poor of summation, picnum_Gop is the frame number that an image sets comprises image, and NumberofCodedPFrame is the frame number of all P two field pictures of having encoded of present image group;
H. according to the target bit of the current frame image that calculates in the step g, first percent of pass distortion model
Calculate model parameter C, wherein R is the actual bit number that last P two field picture takies, and X is the actual complex degree of last P two field picture, and QP is the quantization parameter of last P two field picture; Pass through again
Calculate the quantization parameter QP ' of current frame image, wherein X ' is the prediction complexity of current frame image, and its value is the actual complex degree of former frame image, and R ' is the target bit of current frame image; By the quantization parameter of current frame image current frame image is encoded then;
In order to keep the mild of video quality, the variation of the quantization parameter of front and back two field picture is smaller or equal to 2, and promptly the changing value Δ QP of quantization parameter can only be 0,1,2 ,-1 ,-2; Under the prediction complexity of current frame image, corresponding to each Δ QP, there is corresponding Δ R, Δ R is the changing value of the quantization parameter of former frame image image object bit number of correspondence when increasing or reducing; By
Obtain Δ QP and be designated as Δ R respectively for the Δ R value of-2 ,-1,0,1,2 o'clock correspondences
-2, Δ R
-1, Δ R
0, Δ R
1, Δ R
2Because Δ R
-2＞Δ R
-1＞Δ R
0＞Δ R
1＞Δ R
2, again by calculating point1=(Δ R
2+ Δ R
1+ 1)＞＞1; Point2=(Δ R
0+ Δ R
1+ 1)＞＞1; Point3=(Δ R
0+ Δ R
-1+ 1)＞＞1; Point4=(Δ R
-1+ Δ R
-2+ 1) reference point as a comparison＞＞1, symbol "＞＞" for the shift right operation symbol,, reference point is done appropriate adjustment in order to obtain more accurate quantization parameter QP, concrete available following code is realized:
if(DiffBits＜β1*point1)
{QP＝Pre1PFrameQp+2；}
else?if(DiffBits＜β2*point2)
{QP＝Pre1PFrameQp+1；}
else?if(DiffBits＜β3*point3)
{QP＝Pre1PFrameQp；}
else?if(DiffBits＜β4*point4)
{QP＝Pre1PFrameQp-1；}
else
{QP＝Pre1PFrameQp-2；}
QP＝MAX(1，QP＝MIN(63，QP))；
In the above code, DiffBits is the difference of the actual bit number of the target bit FrameBits of current frame image and former frame image, β 1, β 2, β 3, β 4 are for adjusting parameter, general value is { 4,3,2,4}, QP is the quantization parameter of current frame image in the code, and Pre1PFrameQp is the quantization parameter of former frame image, the actual bit number that Pre1PFramebits takies for the former frame image;
If the i.i-1. intact image of present encoding is the I two field picture, then add up the actual bit number that current frame image takies, the remaining bits number of present image group is the poor of the target bit of present image group and actual bit number that current frame image takies, and the initial value that the actual bit that the present image group takies is counted summation is the actual bit number that current frame image takies; If the i-2. intact image of present encoding is the P two field picture, then add up the actual bit number that current frame image takies, calculate the actual complex degree of current frame image, the actual bit number that its value takies for current frame image and the squared difference of corresponding its reference picture of current frame image and product divided by the size of image, when current frame image is first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the initial value of the actual complex degree summation of present image group is the actual complex degree of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted the actual bit that summation and current frame image take and counted sum that the actual bit that the present image group takies is counted summation, and the initial value of the quantization parameter summation of present image group is the quantization parameter of current frame image; When current frame image is not first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the actual complex degree summation of present image group is the actual complex degree summation of the present image group before the current frame image coding and the actual complex degree sum of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted summation and counted sum with the actual bit that current frame image takies that the actual bit that the present image group takies is counted summation, and the quantization parameter summation of present image group is the quantization parameter summation of the preceding present image group of current frame image coding and the quantization parameter sum of current frame image;
If j. the present image group is not that first image sets and current frame image are the P two field pictures, all the average actual bit number that takies of the every two field picture of image encoded group be all the average actual bit rate of image encoded group and the merchant of frame per second, and whether the product of quantization parameter of judging actual bit number that current frame image takies and current frame image is greater than all 2～3 times of product of the average quantisation parameter of image encoded group or less than all the average actual bit number of image encoded group and all 0.4～0.5 times of product of the average quantisation parameter of image encoded group of the average actual bit number that takies of the every frame of image encoded group and all, if, then determine to detect scene and switch through type
Upgrade the remaining bits number of present image group, RemainBits is the remaining bits number of present image group in the formula, FrameReallyBits is the actual bit number that current frame image takies, and avg_bit_rate is all average actual bit rates of image encoded group, and FRAMERATE is a frame per second;
K. judge whether to exist follow-up vision signal, if exist, judge again then whether the image behind the coding is last image of present image group, if, then actual complex degree summation, the quantization parameter summation of the present image group that obtains according to step I and the actual bit that takies are counted summation, calculate the average quantisation parameter of present image group and average actual complex degree, all average actual complex degree, average quantisation parameter and average actual bit rates of image encoded group, jump to step b then, otherwise, execution in step f; If there is no, then finish.
Can solve following two kinds of special circumstances well by step j, first kind of situation is: the complexity of present image group rises to very big suddenly, if do not append corresponding bit number, the front image encoded of present image group almost can consume most of target bit of present image group so, and the not coded image of present image group will cause quality sharply to descend because of target bit is not enough so; Second kind of situation then is: the complexity of present image descends a lot suddenly, do not subtract corresponding bit number if do not chase after, the front image encoded of present image group only consumes the very little part of the target bit of present image group so, the uncoded image of present image group will increase the target bit that distributes greatly so, cause the quality of these images sharply to rise, both of these case all will cause the instability of video quality.Show by experiment: after scene change detection is handled, the mild property of the fine video quality that can obtain.
To adopting the inventive method and adopting the JVT-H017 method on the platform of AVS, under identical experiment condition, compare experiment.
Form cycle tests by coastguard_cif, container_cif, flower_cif, hall_cif, mobile_cif, news_cif, each the 300 frames circulation splicing of eight cycle testss of Sign_Irene_cif, waterfall_cif, get such CIF form cycle tests 9000 frames at different initial target code checks, frame per second is to test under the 30f/s condition, and result data is as shown in table 1:
Table 1 adopts the comprehensive objective performance comparison sheet of the inventive method and JVT-H017 method
Experimental data in the analytical table 1, show: though the actual average code check that the inventive method produces does not have JVT-H017 like that near target bit rate, but also can satisfy the code check requirement for restriction preferably, under the condition of same target code check, the bit number of actual consumption only accounts for about 98% of JVT-H017 method actual consumption bit number, but the video quality of employing the inventive method has on average improved about 0.2dB than the brightness peak signal to noise ratio PSNR (Peak Signal to Noise Ratio) of JVT-H017 method, the big more presentation video quality of its value is good more, mild degree is represented with sample variance VAR () functional value, novel phaneroplasm amount fluctuation is little more for its value, the sample variance value of the inventive method has only about 40% of JVT-H017 method, illustrates that the inventive method is effectively feasible.
In addition, Fig. 3 shows the statistical comparison that adopts the inventive method and the video quality brightness peak signal to noise ratio PSNR that adopts the JVT-H017 method to produce at the initial target code check during for 1.5Mbps, as can be seen from Figure 3 adopt the mild property of video vision of the inventive method better, fluctuation is less.
Claims (1)
1. one kind to real time video signals VBR code rate control method in the AVS cataloged procedure, signal pickup assembly obtains vision signal and described vision signal is carried out the VBR Rate Control in the AVS cataloged procedure encoder from actual scene, described vision signal comprises a plurality of image sets, each image sets that it is characterized in that the AVS coding comprises an I two field picture and at least one P two field picture, and real time video signals VBR code rate control method in the AVS cataloged procedure be may further comprise the steps:
A. the initial quantization parameter of initialization vision signal is set the initial target code check, and obtain the frame per second of encoder, an image sets comprises the frame number of image and the Gao Yukuan of image, the product of the height and width of note image is the size of image;
B. judge whether the residing present image group of current frame image is first image sets of vision signal, if, then the target bit rate of present image group is the initial target code check, the target bit of present image group is that target bit rate and image sets of present image group comprises the frame number of image and the product of the ratio of frame per second, and execution in step f; Otherwise, continue to carry out;
C. according to the average actual complex degree of previous image group and all the average actual complex degree of image encoded group, use formula
Calculate the prediction complexity XGopCurEst of present image group, wherein, XGopPre is the average actual complex degree of previous image group, and XGopAvg is all average actual complex degree of image encoded group;
D. go on foot the prediction complexity XGopCurEst of the present image group that calculates according to c, through type
Calculate the target bit rate GopBitRate of present image group, wherein, XGopCurEst is the prediction complexity of present image group, PrePQpAVG is the average quantisation parameter of previous image group, XGopAvgCurEst is all average complexities of obtaining of the weighted calculation of the prediction complexity of the actual average complexity of image encoded group and present image group, TotalAvgQp is all average quantisation parameter of image encoded group, and avg_bit_rate is all average actual bit rates of image encoded group;
E. according to the target bit rate GopBitRate of the present image group that calculates in the steps d, calculate the target bit of present image group, its value comprises the frame number of image and the product of the ratio of frame per second for target bit rate and an image sets of present image group;
F. judge whether current frame image is I two field picture or first P two field picture of present image group, if, then when the present image group is first image sets, the quantization parameter of I two field picture or first P two field picture is the initial quantization parameter, and by the initial quantization parameter I two field picture or first P two field picture are encoded execution in step i then; When the present image group is not first image sets, the quantization parameter of I two field picture or first P two field picture is the average quantisation parameter of previous image group, and by the average quantisation parameter of previous image group I two field picture or first P two field picture are encoded execution in step i then; Otherwise, continue to carry out;
G. according to the target bit of present image group, through type
Calculate the target bit FrameBits of current frame image, wherein, RemainBits is the remaining bits number of present image group, its value is that the target bit of present image group and the actual bit that the present image group takies are counted the poor of summation, picnum_Gop is the frame number that an image sets comprises image, and NumberofCodedPFrame is the frame number of all P two field pictures of having encoded of present image group;
H. according to the target bit of the current frame image that calculates in the step g, first percent of pass distortion model
Calculate model parameter C, wherein R is the actual bit number that last P two field picture takies, and X is the actual complex degree of last P two field picture, and QP is the quantization parameter of last P two field picture; Pass through again
Calculate the quantization parameter QP ' of current frame image, wherein X ' is the prediction complexity of current frame image, and its value is the actual complex degree of former frame image, and R ' is the target bit of current frame image; By the quantization parameter of current frame image current frame image is encoded then;
If the i.i-1. intact image of present encoding is the I two field picture, then add up the actual bit number that current frame image takies, the remaining bits number of present image group is the poor of the target bit of present image group and actual bit number that current frame image takies, and the initial value that the actual bit that the present image group takies is counted summation is the actual bit number that current frame image takies; If the i-2. intact image of present encoding is the P two field picture, then add up the actual bit number that current frame image takies, calculate the actual complex degree of current frame image, the actual bit number that its value takies for current frame image and the squared difference of corresponding its reference picture of current frame image and product divided by the size of image, when current frame image is first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the initial value of the actual complex degree summation of present image group is the actual complex degree of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted the actual bit that summation and current frame image take and counted sum that the actual bit that the present image group takies is counted summation, and the initial value of the quantization parameter summation of present image group is the quantization parameter of current frame image; When current frame image is not first P two field picture of present image group, the remaining bits number of present image group is the poor of the remaining bits number of the present image group before the current frame image coding and actual bit number that current frame image takies, the actual complex degree summation of present image group is the actual complex degree summation of the present image group before the current frame image coding and the actual complex degree sum of current frame image, it is that actual bit that the present image group before the current frame image coding takies is counted summation and counted sum with the actual bit that current frame image takies that the actual bit that the present image group takies is counted summation, and the quantization parameter summation of present image group is the quantization parameter summation of the preceding present image group of current frame image coding and the quantization parameter sum of current frame image;
If j. the present image group is not that first image sets and current frame image are the P two field pictures, all the average actual bit number that takies of the every two field picture of image encoded group be all the average actual bit rate of image encoded group and the merchant of frame per second, and whether the product of quantization parameter of judging actual bit number that current frame image takies and current frame image is greater than all 2～3 times of product of the average quantisation parameter of image encoded group or less than all the average actual bit number of image encoded group and all 0.4～0.5 times of product of the average quantisation parameter of image encoded group of the average actual bit number that takies of the every frame of image encoded group and all, if, then determine to detect scene and switch through type
Upgrade the remaining bits number of present image group, RemainBits is the remaining bits number of present image group in the formula, FrameReallyBits is the actual bit number that current frame image takies, and avg_bit_rate is all average actual bit rates of image encoded group, and FRAMERATE is a frame per second;
K. judge whether to exist follow-up vision signal, if exist, judge again then whether the image behind the coding is last image of present image group, if, then actual complex degree summation, the quantization parameter summation of the present image group that obtains according to step I and the actual bit that takies are counted summation, calculate the average quantisation parameter of present image group and average actual complex degree, all average actual complex degree, average quantisation parameter and average actual bit rates of image encoded group, jump to step b then, otherwise, execution in step f; If there is no, then finish.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710161158 CN101188755B (en) | 2007-12-14 | 2007-12-14 | A method for VBR code rate control in AVX decoding of real time video signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710161158 CN101188755B (en) | 2007-12-14 | 2007-12-14 | A method for VBR code rate control in AVX decoding of real time video signals |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101188755A CN101188755A (en) | 2008-05-28 |
| CN101188755B true CN101188755B (en) | 2010-06-09 |
Family
ID=39480905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200710161158 Expired - Fee Related CN101188755B (en) | 2007-12-14 | 2007-12-14 | A method for VBR code rate control in AVX decoding of real time video signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101188755B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101677398A (en) * | 2008-09-19 | 2010-03-24 | 三星电子株式会社 | Scene switching code rate control method |
| CN101511026B (en) * | 2009-02-23 | 2011-01-19 | 四川长虹电器股份有限公司 | Code rate control method for AVS secondary encode based on scene |
| CN101931800B (en) * | 2009-06-24 | 2013-08-14 | 财团法人工业技术研究院 | Method and system for encoding region of interest by limited variable bit rate control |
| CN103841418B (en) * | 2012-11-22 | 2016-12-21 | 中国科学院声学研究所 | The optimization method of video monitor Rate Control and system in a kind of 3G network |
| CN104159109B (en) * | 2014-05-28 | 2018-12-11 | 百视通网络电视技术发展有限责任公司 | Bit rate control method and system based on VBR Video coding |
| CN104994387B (en) * | 2015-06-25 | 2017-10-31 | 宁波大学 | A kind of bit rate control method of fused images feature |
| CN107087169B (en) * | 2017-03-22 | 2019-11-12 | 建荣半导体（深圳）有限公司 | A kind of compression method, UVC video equipment and electronic equipment |
| CN108989818B (en) * | 2017-05-31 | 2020-12-08 | 深圳市中兴微电子技术有限公司 | Image coding parameter adjusting method and device |
| CN109413427B (en) * | 2017-08-17 | 2022-04-08 | 腾讯科技（深圳）有限公司 | Video frame coding method and terminal |
| CN108600758A (en) * | 2018-03-26 | 2018-09-28 | 南京地铁建设有限责任公司 | I vertical frame dimensions burst code stream shaping methods based on city rail passenger information system |
| CN110971900A (en) * | 2019-11-14 | 2020-04-07 | 杭州当虹科技股份有限公司 | Code rate control method suitable for 4K and 8K ultrahigh-definition motion flat video |
| CN114598866B (en) * | 2020-12-03 | 2024-05-03 | 北京君正集成电路股份有限公司 | VBR code rate control method for video real-time coding |
| CN113286145B (en) * | 2021-04-26 | 2022-07-22 | 维沃移动通信有限公司 | Video coding method and device and electronic equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1615020A (en) * | 2004-11-10 | 2005-05-11 | 华中科技大学 | Method for pridicting sortable complex in frame |
-
2007
- 2007-12-14 CN CN 200710161158 patent/CN101188755B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1615020A (en) * | 2004-11-10 | 2005-05-11 | 华中科技大学 | Method for pridicting sortable complex in frame |
Non-Patent Citations (2)
| Title |
|---|
| 赵阳生等.应用于AVS视频解码器的VLD设计.微机发展15 9.2005,15(9),122-124. |
| 赵阳生等.应用于AVS视频解码器的VLD设计.微机发展15 9.2005,15(9),122-124. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101188755A (en) | 2008-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101188755B (en) | A method for VBR code rate control in AVX decoding of real time video signals | |
| US8416849B2 (en) | Self-adapting code rate distribution method of video image macro-block level | |
| CN101395923B (en) | Method and apparatus for determining in picture signal encoding the bit allocation for groups of pixel blocks in a picture | |
| CN1910934B (en) | Encoder with adaptive rate control | |
| CN102932641B (en) | A kind of constant-quality bit rate control method | |
| Zhang et al. | Low-complexity and high-efficiency background modeling for surveillance video coding | |
| CN100574447C (en) | Fast intraframe predicting mode selecting method based on the AVS video coding | |
| GB2459671A (en) | Scene Change Detection For Use With Bit-Rate Control Of A Video Compression System | |
| CN108769696A (en) | A kind of DVC-HEVC video transcoding methods based on Fisher discriminates | |
| US20060256858A1 (en) | Method and system for rate control in a video encoder | |
| CN101287123A (en) | Code rate controlling method for video coding based on Rho domain | |
| CN112218084B (en) | High-efficiency video coding standard frame-level code rate control method facing surveillance video | |
| Zhou et al. | Complexity-based intra frame rate control by jointing inter-frame correlation for high efficiency video coding | |
| KR101868270B1 (en) | Content-aware video encoding method, controller and system based on single-pass consistent quality control | |
| CN1457196A (en) | Video encoding method based on prediction time and space domain conerent movement vectors | |
| CN101902642B (en) | Quick decision method for H.264 interframe SKIP modes | |
| Luo et al. | A new algorithm on MPEG-2 target bit-number allocation at scene changes | |
| Vu et al. | VMAF based quantization parameter prediction model for low resolution video coding | |
| CN104796704A (en) | Microblock rate control method used for scalable video coding | |
| KR100316764B1 (en) | Method and system for coding images using human visual sensitivity characteristic | |
| CN105681784B (en) | A kind of PSNR blind estimating methods based on H264/AVC videos | |
| CN114554224B (en) | Block-level Lagrange multiplier optimization method based on core reference frame | |
| CN115633176B (en) | Method, device and storage medium for dynamically adjusting length of picture group | |
| US20110182343A1 (en) | Encoder | |
| CN117979006B (en) | Low-illumination live video real-time transcoding enhancement bright structure quantization method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100609 Termination date: 20131214 |