CN107257464A - HD video encoder bit rate control algolithm based on Sobel operators and linear regression - Google Patents
HD video encoder bit rate control algolithm based on Sobel operators and linear regression Download PDFInfo
- Publication number
- CN107257464A CN107257464A CN201611246695.7A CN201611246695A CN107257464A CN 107257464 A CN107257464 A CN 107257464A CN 201611246695 A CN201611246695 A CN 201611246695A CN 107257464 A CN107257464 A CN 107257464A
- Authority
- CN
- China
- Prior art keywords
- frame
- linear regression
- lcu
- gradient
- sobel operators
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
- H04N19/149—Data rate or code amount at the encoder output by estimating the code amount by means of a model, e.g. mathematical model or statistical model
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/184—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being bits, e.g. of the compressed video stream
Abstract
The present invention provides a kind of HD video encoder bit rate control algolithm based on Sobel operators and linear regression for HEVC video encoding standards.Main the first frame using the linear regression model (LRM) learnt for coding of the invention finds optimal initial QP (quantization parameter), so that cataloged procedure rapidly adapts to the bandwidth of setting, and the First-order Gradient information of frame in and interframe is produced using Sobel operators, then adaptively selected Optimal gradient information substitution encoder complexity turns into the weight of LCU layers of bit distribution.The inventive method makes bit rate error smaller, and bit rate output is more steady, and the subjective and objective quality of video is more preferable.Test result indicates that, the inventive method is compared with HEVC standard Rate Control motion K0103, and bit rate error averagely reduces about 0.67%, and Y-PSNR averagely improves about 0.44dB.
Description
Technical field
The present invention relates to the video frequency coding rate control problem in field of picture communication, regarded more particularly, to high definition of new generation
Frequency coding standard HEVC interframe rate control algorithm.
Background technology
In recent years, as high definition, ultra high-definition video are continued to bring out, original video encoding standard H.264/AVC by
It gradually can not meet requirement of the people for video quality.In order to solve this problem, VCEG and MPEG have set up Video coding connection
Conjunction group (JCT-VC), has jointly formulated video encoding standard of new generation --- HEVC.It is new due to being added in many modules
Coding techniques, HEVC is realized on the basis of same video quality is ensured, compared to H.264, and compression efficiency is improved into one
Times.
Although HEVC compression efficiency is greatly improved, in the environment of existing real-time transmission of video, channel width is still difficult
To be met.In order in the case of Bandwidth-Constrained, it is ensured that the transmission quality of video, and time delay is not caused, it would be desirable to
Rate Control is carried out to the cataloged procedure of video.So-called Rate Control, is exactly pre- by the distribution of rational bit and accurate QP
Survey so that the bit rate after Video coding meets required rate limit, and coding distortion is also small as far as possible.
HEVC successively employs three kinds of code control models:R-Q models, R- ρ models, R-lambda models.R-Q models are for code
There is famous " laying hen antinomy " when quantization step is calculated not comprising head bit in rate R calculating.R- ρ moulds
Type is applied to the transform block of fixed size, and it is variable that HEVC, which encodes used transform block size,.R-lambda model energy
It is enough that quantization parameter QP is determined before rate-distortion optimization process, encoder complexity is greatly reduced, and a bit is contained, it is not required to
Consider transform block size problem, it is more outstanding than R-Q models and R- ρ models.Current HEVC standard code uses R-
Lambda models.
At present on how to make actual bit more steady and subjective quality is more preferable closer to target bits, bit rate output
Innovatory algorithm is a lot.Huiling Zhao et al. are proposed to be used in LCU layers of bit allocation procedures of HEVC Rate Controls
SSIM (structural similarity), which substitutes encoder complexity, turns into the scheme of distribution weight, is obtained more test result indicates that suggesting plans
Good subjective quality effect.Jiangtao Wen et al. are proposed before the bit distribution of HEVC Rate Controls, use precoding
16 × 16 CU (coding unit) block message go the complexity information of 64 × 64LCU of estimation (maximum coding unit) block so that
The weight distributed as bit, test result indicates that, compared with canonical algorithm, the bit rate error of institute's extracting method is smaller.
MiaohuiWang et al. proposes the improved HEVC frame in Rate Control bit allocation schemes based on content frame complexity, i.e.,
Target bits are distributed according to the content frame complexity weighed with gradient, test result indicates that, suggest plans and compare canonical algorithm
Bit rate error it is smaller, bit rate output is more steady and subjective and objective quality is more preferable.
The content of the invention
The band of setting can not be rapidly adapted to for presence in video compression coding standard HEVC rate control process of new generation
Wide and encoder complexity and the problem of HVS (human visual system) inconsistent, the present invention propose it is a kind of based on Sobel operators with
The HD video encoder bit rate control algolithm of linear regression, significantly reduces bit rate error, effectively reduces time delay, and improve
The Subjective and objective qualities of video, have a good application prospect in the real-time coding transmitting procedure of Bandwidth-Constrained.
The basic thought of the present invention:1st, the first frame using the linear regression model (LRM) learnt for coding finds optimal initial
QP (quantization parameter), so that cataloged procedure rapidly adapts to the bandwidth of setting.2nd, frame in and interframe are produced using Sobel operators
First-order Gradient information, the weight that then adaptively selected Optimal gradient information substitution encoder complexity is distributed as LCU layers of bit.
The present invention provides a kind of HD video based on Sobel operators and linear regression for HEVC video encoding standards
Encoder bit rate control algolithm.The algorithm mainly includes optimal initial QP determination and the improvement of LCU layers of bit allocation scheme.For
First frame, exports optimal initial QP using the linear regression model (LRM) learnt, to the Rate Control of frame-layer, makes cataloged procedure fast
Speed adapts to the bandwidth of setting.In LCU layers of bit distribution, the First-order Gradient information of frame in and interframe is produced using Sobel operators,
For I frames, directly using frame in gradient as Optimal gradient information, for non-I frames, using the smaller value of frame in and interframe gradient
As Optimal gradient information, the bit of distribution is then determined according to weights of the current LCU in the uncoded LCU of present frame.Specifically
Mainly include procedure below step:
(1) average target bit, average gradient and corresponding optimum for collecting standard video sequence the first frame pixel are initial
QP is normalized as training set, and to training set;
(2) it is put into linear regression model (LRM) to be learned to be trained, obtains the linear regression model (LRM) learnt;
(3) according to the target bit rate of setting, frame per second, GOP sizes and encoded actual bit, GOP layers of bit are distributed;
(4) place GOP weight and buffer state is accounted for according to present frame, frame-layer bit is distributed;
(5) whether be first frame, if the first frame if judging current encoded frame, utilizes distributed frame-layer bit to obtain first
The average target bit of frame pixel, the average frame inside gradient of the first frame pixel is obtained using Sobel operators, carries out normalizing
Change, then input the linear regression model (LRM) learnt, model output optimal initial QP, directly the code check control to carry out frame-layer
System.If not the first frame, then perform step (6);
(6) whether be I frame, if I frames if judging current encoded frame, by the use of Sobel operators produce LCU frame in gradient as
Optimal gradient information, and present frame is traveled through, it is cumulative to obtain frame-layer total gradient.If not I frames, produce LCU's using Sobel operators
Frame in and interframe gradient, take both smaller values as Optimal gradient information, and travel through present frame, and adding up, it is always terraced to obtain frame-layer
Degree;
(7) weight of the uncoded LCU gradients sum of present frame is accounted for according to current LCU gradients, target bits are distributed, and carry out
Actual coding;
(8) parameter renewal is carried out according to the deviation of current LCU targets and actual bit.Judge whether to travel through present frame, if
Do not travel through, then perform step (7).If having traveled through, step (9) is performed;
(9) according to the deviation adjusting buffer state of present frame target and actual bit.Judge whether to travel through current GOP
(image sets), if not traveling through, perform step (4).If having traveled through, step (10) is performed;
(10) judge whether encoded completion, if not completing, perform step (3).If having completed, step is performed
(11);
(11) terminate.
In the above-mentioned technical proposal of the present invention, the linear regression model (LRM) is the average criterion of the first frame pixel of description
Relational model between bit, average gradient and corresponding optimum initial Q P, i.e., input average criterion ratio into linear regression model (LRM)
Special and average gradient, the optimal initial QP of model output prediction.The specific formula for calculation of linear regression model (LRM) is:
QPInitialBest=9.4973gradave-23.0407·bpptar+29.965 (1)
Wherein, bpptarThe average number of bits obtained for each pixel, gradaveFor being averaged for each pixel of the first frame
Grad, QPInitialBestFor the optimal initial QP of prediction.
It is described that training set is normalized to carry out the data in training set in the above-mentioned technical proposal of the present invention
Equal proportion is scaled, and data are all zoomed in the range of [0,100].Normalized specific formula for calculation is:
Wherein, XnormFor the data after normalization, X is original data, XmaxAnd XminRespectively raw data set is most
Big value and minimum value.
In the above-mentioned technical proposal of the present invention, the frame in gradient is current LCU each pixel warps in addition to marginal point
Cross the calculating of Sobel operators and obtain the cumulative of brightness value, specific formula for calculation is:
Wherein, ShFor the summation of LCU transverse gradients values of all pixels point in addition to marginal point, SvFor a LCU flash trimming
The summation of longitudinal Grad of the outer all pixels point of edge point, Gintra(i, j, k) is kth frame with the LCU frame ins that (i, j) is starting point
Grad, M and N are respectively the length and width of video sequence.
The present invention above-mentioned technical proposal in, the interframe gradient be current LCU in addition to marginal point each pixel and
Luminance difference between reference frame same position pixel calculates the cumulative of obtained value by Sobel operators, specific to calculate public
Formula is:
R (x, y)=If(x,y)-Ir(x,y) (6)
Wherein, If(x, y) is the brightness value of current pixel point, Ir(x, y) is the brightness of corresponding pixel in reference frame
Value, R (x, y) is current pixel point and the difference of reference pixel point brightness value, Ginter(i, j, k) be kth frame with (i, j) for starting point
LCU interframe Grad.
In the above-mentioned technical proposal of the present invention, the Optimal gradient information of the non-I frames LCU is to utilize Sobel operator meters
Obtained frame in and the smaller value of interframe gradient.
In the above-mentioned technical proposal of the present invention, it is uncoded that the weight of the current LCU accounts for present frame for current LCU gradients
The ratio of LCU gradient sums.
The above-mentioned HD video based on Sobel operators and linear regression of execution can be worked out according to the above method of the present invention
The HEVC video encoders of encoder bit rate control algolithm.
The present invention is to be analyzed and completed based on following thinking:
The initial Q P of first frame has very big influence to bandwidth, on the one hand, initial Q P, which crosses conference, reduces the first frame
PSNR, wastes bandwidth, and on the other hand, initial Q P is too small and may be such that subsequent frame is buffered when being encoded
Situations such as area's overflow, frame-skipping, significantly impact the performance of code control, it is therefore desirable to select a suitable initial Q P.And standard
HEVC rate control algorithms are for the different video sequence of equal resolution, and the initial Q P used when target bit rate is identical is one
Sample, this obviously can exert an adverse impact to the performance of code control, because different video sequences has different contents complicated
Degree can just make cataloged procedure rapidly adapt to the bandwidth set, it is necessary to encoded with different initial Q P.It is theoretical it has been proved that
Linear regression can be determined complementary quantitative between two or more variable using the regression analysis in mathematical statistics
Relation.Therefore, the present invention is by the way that the average target bit and average gradient of the first frame pixel of cycle tests to be put into and learn
Linear regression model (LRM) is practised, optimal initial QP can be adaptively exported, so that cataloged procedure rapidly adapts to the bandwidth of setting.
Standard HEVC rate control algorithms are when LCU layers of bit are distributed, using the MAD of corresponding LCU in reference frame
(absolute mean error) characterizes encoder complexity, and it does not consider HVS (human visual system) between encoder complexity
The high region of the problem of having inconsistent, i.e. encoder complexity, human eye may be to its distortion and insensitive, so as to cause bit
Waste and bit rate error the problems such as become big, and when scene switches, corresponding LCU MAD can not in reference frame
For characterizing current LCU encoder complexity, PSNR decline result in.Theory is it has been proved that gradient information can be well
Complexity information is characterized, and is consistent with HVS.Sobel operators not only have smoothing effect to noise, and can be in image
Any point produce corresponding gradient vector, therefore the present invention obtains the gradient information of pixel using Sobel operators, from
And LCU layers of bit are distributed by the use of gradient information as weight.
Compared with the HEVC code rate controlling method for video coding of standard, method bit rate error of the invention is smaller, output
Code check is more steady and subjective and objective quality is more preferable.The method of the present invention calculates optimal initial using linear regression model (LRM) adaptometer
QP so that cataloged procedure rapidly adapts to the bandwidth of setting, so as to reduce bit rate error.The method of the present invention is utilized and is based on
The gradient information of Sobel operators, which substitutes encoder complexity, turns into the weight of LCU layers of bit distribution, in the sensitive region of human eye distortion
More bits are assigned with, so as to effectively increase the subjective and objective quality of video.By the inventive method, ratio can be greatly decreased
Special rate error, makes coding output bit more steady, while improving video subjective and objective quality.
Brief description of the drawings
Fig. 1 is the flow of the HD video encoder bit rate control algolithm based on Sobel operators and linear regression of the present invention
Figure.
Fig. 2~Fig. 3 is the inventive method and the steady comparison diagram of code check of the motion of standard K 0103, and wherein Fig. 2 is sequence
The actual bit contrast of every frame codings of the BlowingBubbles when target bit rate is 3603Kbps;Fig. 3 is sequence
The actual bit contrast of every frame codings of the BQSquare when target bit rate is 2584Kbps.
Fig. 4~Fig. 5 is the inventive method and the distortion performance comparison diagram of the motion of standard K 0103, and wherein Fig. 4 is sequence
BQMall rate distortion curve;Fig. 5 is the rate distortion curve of Kimono1 sequences.
Fig. 6~Fig. 7 is the inventive method and the subjective quality comparison diagram of the motion of standard K 0103, and wherein Fig. 6 is standard
197th frame reconstructed images of the sequence B QMall that K0103 motions are obtained when target bit rate is 10007kbps;Fig. 7 is this hair
197th frame reconstructed images of the sequence B QMall that bright method is obtained when target bit rate is 10007kbps.
Embodiment
With reference to embodiment, the present invention is described in further detail, it is necessary to, it is noted that following embodiment
It is served only for that the present invention is described further, it is impossible to be interpreted as limiting the scope of the invention, art technology is ripe
Personnel are known according to foregoing invention content, some nonessential modifications and adaptations are made to the present invention and are embodied, should still be belonged to
In protection scope of the present invention.
The HD video encoder bit rate control algolithm based on Sobel operators and linear regression of the present invention, with HEVC standard
Identifying code HM10.0 inter-frame encoding methods comparison procedure is as follows:
1st, standard HM10.0 identifying codes are opened, configuration file is lowdelay_P_main, will not use Rate Control side
Method encode obtained bit rate output as the target bit rate of the comparison procedure.
2nd, the inventive method is compared the bit rate control method K0103 motions with standard HEVC identifying codes HM10.0
Compared with.The inventive method and the program of standard method are opened simultaneously, set identical configuration file, it is accordingly required in particular to note, it is necessary to
Rate Control and initial Q P settings are opened, cataloged procedure is then carried out.To four kinds of video coding performances:Bit rate error, PSNR
(wherein bit rate error represents the essence of Rate Control for (Y-PSNR), every frame actual coding output bit and subjective quality
Degree, PSNR embodies the objective quality of encoded video, and per frame actual coding, output bit represents the smoothness of code check) it is compared point
Analysis, the gap for comparing performance is evaluated with three below index:
Δ PSNR=PSNRproposed-PSNRK0103 (11)
Wherein, BRactFor actual bit rate, BRtarFor target bit rate, PSNRproposedThe peak value letter obtained for the inventive method
Make an uproar and compare, PSNRK0103The Y-PSNR obtained for the motion of standard K 0103, Δ BRerrorCarried for the inventive method with standard K 0103
The actual bit rate of case and the code check deviation percentage of target bit rate, Δ PSNR are the inventive method and the peak of the motion of standard K 0103
It is worth the difference of signal to noise ratio.
3rd, coded object is HEVC standard video sequence, and their title, resolution ratio and frame per second is respectively:
BlowingBubbles (416x240,50 frames/second), BQSquare (416x240,60 frames/second), Flowervase (416x240,
30 frames/second) and BQMall (832x480,60 frames/second), RaceHorses (832x480,30 frames/second), BasketballDrill
(832x480,50 frames/second) and Vidyo1 (1280x720,60 frames/second), FourPeople (1280x720,60 frames/second),
Vidyo4 (1280x720,60 frames/second) and Kimono1 (1920x1080,24 frames/second), Cactus (1920x1080,50 frames/
Second), BQTerrace (1920x1080,60 frames/second).
4th, 2 identical video test sequences are inputted;
5th, Video coding is carried out in HM10.0 identifying codes to standard video sequence using the motion of standard K 0103;
6th, Video coding is carried out in HM10.0 identifying codes to standard video sequence using the inventive method;
7th, two programs export the code check after Video coding and PSNR, result such as 1~table of table 2 of above-mentioned 2 indexs respectively
It is shown.Test result indicates that, the inventive method and the bit rate error of the motion of standard K 0103 averagely reduce about 0.67%, peak value
Signal to noise ratio averagely improves about 0.44dB.This experiment is tested using 12 standard video sequences of different resolution, the present invention
The above-mentioned two index of method is above standard K0103 motions, fully demonstrates the universality of the inventive method.
The inventive method of table 1 and the bit rate error of the motion of standard K 0103 are contrasted
The inventive method of table 2 and the PSNR values of the motion of standard K 0103 are contrasted
Claims (8)
1. a kind of HD video encoder bit rate control algolithm based on Sobel operators and linear regression, is regarded primarily directed to HEVC
Improve Rate Control part in frequency coding standard.It is primarily characterized in that including procedure below step:
(1) average target bit, average gradient and corresponding optimum initial Q P for collecting standard video sequence the first frame pixel are made
For training set, and training set is normalized;
(2) it is put into linear regression model (LRM) to be learned to be trained, obtains the linear regression model (LRM) learnt;
(3) according to the target bit rate of setting, frame per second, GOP sizes and encoded actual bit, GOP layers of bit are distributed;
(4) place GOP weight and buffer state is accounted for according to present frame, frame-layer bit is distributed;
(5) whether be first frame, if the first frame if judging current encoded frame, utilizes distributed frame-layer bit to obtain the first frame picture
The average target bit of vegetarian refreshments, the average frame inside gradient of the first frame pixel is obtained using Sobel operators, is normalized, so
The linear regression model (LRM) learnt, directly model output optimal initial QP, the Rate Control to carry out frame-layer are inputted afterwards.If no
It is the first frame, then performs step (6);
(6) whether be I frame, if I frames if judging current encoded frame, LCU frame in gradient is produced as optimal by the use of Sobel operators
Gradient information, and present frame is traveled through, it is cumulative to obtain frame-layer total gradient.If not I frames, LCU frame in is produced using Sobel operators
With interframe gradient, both smaller values are taken as Optimal gradient information, and travel through present frame, it is cumulative to obtain frame-layer total gradient;
(7) weight of the uncoded LCU gradients sum of present frame is accounted for according to current LCU gradients, target bits are distributed, and carry out reality
Coding;
(8) parameter renewal is carried out according to the deviation of current LCU targets and actual bit.Judge whether to travel through present frame, if not time
Go through, then perform step (7).If having traveled through, step (9) is performed;
(9) according to the deviation adjusting buffer state of present frame target and actual bit.Judge whether to travel through current GOP (images
Group), if not traveling through, perform step (4).If having traveled through, step (10) is performed;
(10) judge whether encoded completion, if not completing, perform step (3).If having completed, step (11) is performed;
(11) terminate.
2. the HD video encoder bit rate control algolithm as claimed in claim 1 based on Sobel operators and linear regression, it is special
Levy be linear regression model (LRM) for description the first frame pixel average target bit, average gradient and corresponding optimum initial Q P it
Between relational model, i.e., input average target bit and average gradient into linear regression model (LRM), model output prediction it is optimal
Initial Q P.The specific formula for calculation of linear regression model (LRM) is:
QPInitialBest=9.4973gradave-23.0407bpptar+29.965 (1)
Wherein, bpptarThe average number of bits obtained for each pixel, gradaveFor the average gradient of each pixel of the first frame
Value, QPInitialBestFor the optimal initial QP of prediction.
3. the HD video encoder bit rate control algolithm as claimed in claim 2 based on Sobel operators and linear regression, it is special
Levy and be training set is normalized to carry out equal proportion scaling to the data in training set, data are all zoomed to [0,
100] in the range of.Normalized specific formula for calculation is:
Wherein, XnormFor the data after normalization, X is original data, XmaxAnd XminRespectively the maximum of raw data set and
Minimum value.
4. the HD video encoder bit rate control algolithm as claimed in claim 3 based on Sobel operators and linear regression, it is special
Levy and be that each pixel in addition to marginal point obtains the cumulative of brightness value to frame in gradient by the calculating of Sobel operators for current LCU,
Specific formula for calculation is:
Wherein, ShFor the summation of LCU transverse gradients values of all pixels point in addition to marginal point, SvMarginal point is removed for a LCU
The summation of longitudinal Grad of outer all pixels point, Gintra(i, j, k) is kth frame with the LCU frame in gradients that (i, j) is starting point
Value, M and N are respectively the length and width of video sequence.
5. the HD video encoder bit rate control algolithm as claimed in claim 4 based on Sobel operators and linear regression, it is special
Levy and be that interframe gradient is luminance differences of the current LCU in addition to marginal point between each pixel and reference frame same position pixel
Value calculates the cumulative of obtained value by Sobel operators, and specific formula for calculation is:
R (x, y)=If(x,y)-Ir(x,y) (6)
Wherein, If(x, y) is the brightness value of current pixel point, Ir(x, y) is the brightness value of corresponding pixel in reference frame, R
(x, y) is current pixel point and the difference of reference pixel point brightness value, Ginter(i, j, k) is that kth frame is starting point with (i, j)
LCU interframe Grad.
6. being calculated based on Sobel operators and the control of the HD video encoder bit rate of linear regression as described in one of claim 4~5
Method, it is characterised in that non-I frames LCU Optimal gradient information is to calculate obtained frame in and interframe gradient using Sobel operators
Smaller value.
7. the HD video encoder bit rate control algolithm as claimed in claim 6 based on Sobel operators and linear regression, it is special
Levy the ratio for being that current LCU weight accounts for the uncoded LCU gradients sum of present frame for current LCU gradients.
8. a kind of be used for described in perform claim requirement one of 1~7 based on Sobel operators and the HD video of linear regression coding code
The HEVC video encoders of rate control algolithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611246695.7A CN107257464B (en) | 2016-12-29 | 2016-12-29 | High-definition video coding rate control method based on Sobel operator and linear regression |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611246695.7A CN107257464B (en) | 2016-12-29 | 2016-12-29 | High-definition video coding rate control method based on Sobel operator and linear regression |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107257464A true CN107257464A (en) | 2017-10-17 |
CN107257464B CN107257464B (en) | 2020-04-03 |
Family
ID=60027132
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611246695.7A Active CN107257464B (en) | 2016-12-29 | 2016-12-29 | High-definition video coding rate control method based on Sobel operator and linear regression |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107257464B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110267045A (en) * | 2019-08-07 | 2019-09-20 | 杭州微帧信息科技有限公司 | A kind of method, apparatus and readable storage medium storing program for executing that video is handled and encoded |
CN110572653A (en) * | 2019-09-27 | 2019-12-13 | 腾讯科技(深圳)有限公司 | Video encoding method, video decoding method, video encoding apparatus, video decoding apparatus, storage medium, and electronic apparatus |
CN110677647A (en) * | 2019-09-27 | 2020-01-10 | 腾讯科技(深圳)有限公司 | Video decoding method, video encoding method, video decoding apparatus, video encoding apparatus, storage medium, video decoder, and video encoder |
CN113923454A (en) * | 2021-09-06 | 2022-01-11 | 重庆邮电大学 | Video coding rate control LCU layer bit distribution method and storage medium |
CN114827607A (en) * | 2022-03-25 | 2022-07-29 | 李勤来 | Improved big data video high-fidelity transmission coding regulation and control method |
CN116847101A (en) * | 2023-09-01 | 2023-10-03 | 易方信息科技股份有限公司 | Video bit rate ladder prediction method, system and equipment based on transform network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070223021A1 (en) * | 2006-03-23 | 2007-09-27 | Samsung Electronics Co., Ltd. | Image encoding/decoding method and apparatus |
CN101572806A (en) * | 2009-06-01 | 2009-11-04 | 北京邮电大学 | Frame I code rate control method based on H264 |
CN102930268A (en) * | 2012-08-31 | 2013-02-13 | 西北工业大学 | Accurate positioning method for data matrix code under pollution and multi-view situation |
-
2016
- 2016-12-29 CN CN201611246695.7A patent/CN107257464B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070223021A1 (en) * | 2006-03-23 | 2007-09-27 | Samsung Electronics Co., Ltd. | Image encoding/decoding method and apparatus |
CN101572806A (en) * | 2009-06-01 | 2009-11-04 | 北京邮电大学 | Frame I code rate control method based on H264 |
CN102930268A (en) * | 2012-08-31 | 2013-02-13 | 西北工业大学 | Accurate positioning method for data matrix code under pollution and multi-view situation |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110267045A (en) * | 2019-08-07 | 2019-09-20 | 杭州微帧信息科技有限公司 | A kind of method, apparatus and readable storage medium storing program for executing that video is handled and encoded |
CN110267045B (en) * | 2019-08-07 | 2021-09-24 | 杭州微帧信息科技有限公司 | Video processing and encoding method, device and readable storage medium |
CN110572653A (en) * | 2019-09-27 | 2019-12-13 | 腾讯科技(深圳)有限公司 | Video encoding method, video decoding method, video encoding apparatus, video decoding apparatus, storage medium, and electronic apparatus |
CN110677647A (en) * | 2019-09-27 | 2020-01-10 | 腾讯科技(深圳)有限公司 | Video decoding method, video encoding method, video decoding apparatus, video encoding apparatus, storage medium, video decoder, and video encoder |
CN110677647B (en) * | 2019-09-27 | 2024-02-09 | 腾讯科技(深圳)有限公司 | Video decoding and encoding methods and devices, storage medium, decoder and encoder |
CN113923454A (en) * | 2021-09-06 | 2022-01-11 | 重庆邮电大学 | Video coding rate control LCU layer bit distribution method and storage medium |
CN113923454B (en) * | 2021-09-06 | 2024-01-12 | 北京盛大博通文化发展有限公司 | Video coding rate control LCU layer bit allocation method and storage medium |
CN114827607A (en) * | 2022-03-25 | 2022-07-29 | 李勤来 | Improved big data video high-fidelity transmission coding regulation and control method |
CN116847101A (en) * | 2023-09-01 | 2023-10-03 | 易方信息科技股份有限公司 | Video bit rate ladder prediction method, system and equipment based on transform network |
CN116847101B (en) * | 2023-09-01 | 2024-02-13 | 易方信息科技股份有限公司 | Video bit rate ladder prediction method, system and equipment based on transform network |
Also Published As
Publication number | Publication date |
---|---|
CN107257464B (en) | 2020-04-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107257464A (en) | HD video encoder bit rate control algolithm based on Sobel operators and linear regression | |
CN101710995B (en) | Video coding system based on vision characteristic | |
CN106534862B (en) | Video coding method | |
EP2227022A1 (en) | A self-adapting code rate distribution method of video image macro-block level | |
CN104796705B (en) | A kind of HEVC Video codings rate-distortion optimization and rate control algorithm based on SSIM | |
CN108989802B (en) | HEVC video stream quality estimation method and system by utilizing inter-frame relation | |
CN108063944B (en) | Perception code rate control method based on visual saliency | |
CN104185024B (en) | A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model | |
CN102137258B (en) | Method for controlling three-dimensional video code rates | |
CN105120290B (en) | A kind of deep video fast encoding method | |
CN103873861A (en) | Coding mode selection method for HEVC (high efficiency video coding) | |
CN103634601B (en) | Structural similarity-based efficient video code perceiving code rate control optimizing method | |
CN105681797B (en) | A kind of DVC-HEVC video transcoding methods based on prediction residual | |
CN102547225B (en) | Video monitoring scene judgment method and device, and monitored image encoding method and device | |
CN106791848B (en) | Two-Pass code rate control method based on HEVC | |
CN105187826B (en) | For the fast intra mode decision method of high efficiency video encoding standard | |
EP3545677A1 (en) | Methods and apparatuses for encoding and decoding video based on perceptual metric classification | |
CN112825557B (en) | Self-adaptive sensing time-space domain quantization method aiming at video coding | |
CN111447446B (en) | HEVC (high efficiency video coding) rate control method based on human eye visual region importance analysis | |
CN105578181A (en) | Rapid intra-frame mode decision and block matching method for screen content compression in HEVC (High Efficiency Video Coding) | |
CN108769696A (en) | A kind of DVC-HEVC video transcoding methods based on Fisher discriminates | |
CN106791828A (en) | High performance video code-transferring method and its transcoder based on machine learning | |
CN102752591B (en) | H.264 code rate control method based on comprehensive factor | |
CN106534855B (en) | A kind of Lagrange factor calculation method towards SATD | |
CN100442848C (en) | Method for controlling code rate in H263 coding |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |