CN105208389A - Video encoding hard decision quantization method based on content self-adaption deviation model - Google Patents
Video encoding hard decision quantization method based on content self-adaption deviation model Download PDFInfo
- Publication number
- CN105208389A CN105208389A CN201510614662.2A CN201510614662A CN105208389A CN 105208389 A CN105208389 A CN 105208389A CN 201510614662 A CN201510614662 A CN 201510614662A CN 105208389 A CN105208389 A CN 105208389A
- Authority
- CN
- China
- Prior art keywords
- quantization
- algorithm
- play amount
- model
- hard decision
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a video encoding dead zone quantization method based on content self-adaption. A content self-adaption deviation amount model is adopted, and a content self-adaption quantization algorithm is achieved. The model can achieve self-adaption changes of the deviation amount according to different video contents, and meanwhile, obvious improvement is achieved in the aspect of rate-distortion performance compared with a dead zone HDQ algorithm with the fixed deviation amount. Because hard decision quantization with independent coefficients is adopted for the algorithm, quantization performs parallel processing at the coefficient level and can be easily achieved on hardware. Independent statistics and analysis are performed based on a DCT coefficient, accurate distribution parameters are obtained, and the coefficient self-adaption quantization deviation amount model is constructed through the distribution parameters and quantization parameters. During video encoding quantization, self-adapting dead zone HDQ quantization based on the model is adopted, and efficient video encoding is achieved.
Description
Technical field
Algorithm of the present invention is applicable to the quantiser design of H.264 video encoder, is equally applicable to the quantiser design of H.265/HEVC video encoder, specifically a kind of Video coding hard decision quantization method of content-based self adaptation skew model.
Background technology
The nucleus module that quantification is MPEG and H.26x etc. damages in video compression encoder, the conversion coefficient compared with great dynamic range is quantized to be mapped to limited quantized interval, make full use of the perception redundancy properties of human eye vision anamorphic system, in less perceptual distortion situation, realize Signal Compression.Quantize to decide video compression distortion size, also have great impact to Rate Control.Define quantization step parameter (Qp) in video encoding and decoding standard, and define the details of analog quantity, details is not realized to the quantification of conversion coefficient and make concrete regulation.In Video Codec in early days, conversion coefficient have employed uniform scalar quantization (uniformscalarquantizer, USQ).Afterwards, H.264/AVC waited in correlative coding standard with early stage the hard decision quantization algorithm (Hard-decisionquantization, HDQ) that have employed based on simply rounding up at MPEG-4.Scholar is had again to propose afterwards, the dead band hard decision of constant offset amount is adopted to quantize (deadzoneHDQ) algorithm, soft-decision quantization algorithm (soft-decisionquantizationSDQ), and simplify version SDQ algorithm, i.e. rate-distortion optimization quantization algorithm (RatedistortionoptimizationquantizationRDOQ).
Above-mentioned quantization algorithm, respectively has pluses and minuses, and the present invention takes full advantage of the advantage in above-mentioned algorithm, makes up again the segmental defect in above-mentioned algorithm simultaneously.
Immediate prior art 1: the dead band HDQ algorithm of self adaptation constant offset amount.
This algorithm utilizes the statistical property of entropy code, have employed different dead-zone quantization side-play amounts respectively to intra-frame prediction block and interframe prediction block.Experimental verification, it is 1/3 and 1/6 more suitable that the side-play amount of intra-frame prediction block and interframe prediction block is got respectively.Compare in early stage quantizer the HDQ algorithm that simply rounds up, the dead band HDQ algorithm of employing constant offset amount can larger enhancing rate distortion performance.H.264/AVC and the associated encoder of HEVC standard all adopt this algorithm.
The defect of this prior art:
This algorithm still belongs to HDQ algorithm, thinks that adjacent coefficient is separate in block, namely thinks that signal source is memoryless signal source.But based in contextual entropy code algorithm, the hypothesis of memoryless signal source is invalid.So in distortion performance, this algorithm is not optimum.
Immediate prior art 2: soft-decision quantizes (SDQ) algorithm
SDQ algorithm adopts HDQ algorithm to carry out pre-quantization, by all integers alternatively quantized value between zero to pre-quantization value, utilizes Viterbi all direction search method, selects optimal quantization value under rate-distortion optimization criterion.This optimum choice carries out quantification to coefficients all in transform block simultaneously and realizes.Relative pre-quantization result, this algorithm is finely tuned the quantized result of part coefficient in block, considers influencing each other between rate distortion Coding cost and coefficient, performance has very large lifting compared to HDQ during adjustment.
The defect of this prior art:
This algorithm takes into full account the correlation between coefficient, have employed Viterbi all direction search method simultaneously, though performance is excellent in performance, but, the computation complexity brought by Viterbi search is very high, and the correlation between coefficient and serial seriously hinder the effective realization of SDQ algorithm on hardware.
Immediate prior art 3:RDOQ algorithm
RDOQ algorithm is on SDQ algorithm basis, for reducing the shortcut calculation that SDQ algorithm complex proposes.RDOQ algorithm adopts HDQ algorithm to carry out pre-quantization equally, but this algorithm generally only chooses 3 possible candidate quantisation values; Replace whole mesh search with local path search, thus reduce computation complexity widely.Performance still can obtain the gain of most of SDQ algorithm coding.At present, H.264/AVC the JM identifying code of standard and the HM identifying code of HEVC standard all have employed RDOQ algorithm.
The defect of this prior art: this algorithm is the simplification to SDQ algorithm, but mainly for the video encoder based on software simulating, and fail to eliminate the serial caused by viterbi algorithm branching selection and contextual arithmetic.Therefore, high between data correlation still hinders the realization of this algorithm on hardware.
Summary of the invention
The present invention proposes a kind of content-based adaptive Video coding dead-zone quantization method, adopts content-adaptive to quantize side-play amount model, realizes content-adaptive quantization algorithm.This model according to the difference of video content, can realize the adaptive change quantizing side-play amount.Meanwhile, in distortion performance, the dead band HDQ algorithm compared to constant offset amount is significantly increased; Because algorithm adopts coefficient independently hard decision quantification, quantification can carry out parallel processing in coefficient level, hardware is easy to realize.The present invention is based on the analysis of DCT coefficient independent statistics, obtain distributed constant more accurately, utilize distributed constant and quantization parameter to build coefficient adaptive quantification side-play amount model.When Video coding quantizes, adopt the adaptive dead zone HDQ based on this model to quantize, realize efficient video coding.
The technical problem that the present invention need solve:
(1) based on the dead band hard decision quantization method that adaptive quantizing side-play amount model, low serial rely on;
(2) based on just sentencing maximization, probability of miscarriage of justice minimized self adaptation side-play amount model building method;
(3) functional relation between optimized migration amount model and DCT distributed constant and quantization parameter how is determined;
The present invention proposes following technical scheme for this reason:
(1) the present invention proposes a kind of low serial based on adaptive quantizing side-play amount model and relies on dead band hard decision quantization method, the result of (pre-quantization) is quantized according to the hard decision that simply rounds up, determine that 2-3 may preferably candidate quantisation result, based on the coefficient level adaptation side-play amount model that the present invention proposes, do not producing between coefficient under serial dependence prerequisite, adjustment is being carried out to pre-quantization result and realizes quantizing to optimize;
(2) propose off-line statistical analysis, build self adaptation side-play amount model: contrast round up hard decision pre-quantization and SDQ quantized result, if both are different, exploration makes, between the possible offset field that hard decision quantizes and SDQ comes to the same thing, to be called between forward region; If both are identical, sound out and make hard decision quantification and SDQ result keep, between identical offset field, being called inverse direction intervals; Based on just sentencing maximization, probability of miscarriage of justice minimum restriction, getting between two kinds of offset field and occuring simultaneously, and adopting block diagram statistical analysis technique, to the positive and negative side-play amount span fractional analysis of all coefficients, determine each coefficient optimized migration amount;
(3) by side-play amount model construction be the function of DCT coefficient distributed constant and quantization parameter, (2) method of employing gathers the sample in different sequence and different target code check situation in a large number, adopts least square fitting to go out optimum quantization side-play amount model.
The technique effect of this technical scheme
(1) build in self adaptation side-play amount model process, sample collection considers all quantization parameters, takes into account two amounts index contrast result, establishes two parts sample optimized migration weight range separately, seeks common ground.Therefore, the model built based on these samples can improve just sentences probability, reduction probability of miscarriage of justice;
(2) just sentencing under maximization, probability of miscarriage of justice minimum restriction, optimized migration weight range is determined.Adopt block diagram statistical analysis technique, in optimized migration weight range, determine the optimized migration amount of each coefficient.In sample collection procedure, acquire a large amount of different sequence, the sample under different target code check, therefore, by means of the model that great amount of samples builds, the generality of this model is ensured, make the algorithm adopting this model be applicable to different application scenarios, meet the different market demands;
The program proposes new hard decision quantization algorithm, and in quantification judgement, consider the statistical property of DCT coefficient, the self adaptation side-play amount model adopting the present invention to propose, realizes the adjustment of quantization parameter.Therefore, in performance, will have a distinct increment than adopting the dead band hard decision algorithm of constant offset amount; Meanwhile, this algorithm does not produce the serial between coefficient, and does not substantially increase computation complexity.Therefore, this algorithm is a kind of excellent performance, low serial dependence, Video coding quantization algorithm that complexity is low.
Accompanying drawing explanation
Fig. 1 is self adaptation side-play amount modeling procedure figure;
Fig. 2 is I and P/B frame
--Qp statistical value figure mono-;
Fig. 3 is I and P/B frame
--Qp statistical value figure bis-;
Fig. 4 is I/PB frame
--Qp model value figure mono-;
Fig. 5 is I/PB frame
--Qp model value figure bis-;
Fig. 6 is predictive coding procedure chart.
Embodiment
Below in conjunction with accompanying drawing, the invention will be further described.
Have employed prediction differential coding mode based on block image compression coding scheme, this coded system can reduce data correlation over time and space.The data of video compression and coding are the differences of original image and predicted picture.Therefore, the regularity of distribution of residual error data directly will affect coding efficiency.So the regularity of distribution of research residual error data coefficient after dct transform, builds and quantizes the functional relation of side-play amount about distributed constant and quantization parameter.Thus the hard decision quantization algorithm of employing self adaptation side-play amount of the present invention is proposed.
(1) self adaptation side-play amount modeling process
The modeling process of coefficient self adaptation side-play amount comprises data statistics, off-line data fractional analysis and heuristic modeling process.Data statistics comprises counting statistics and the rationally candidate offset quantitative statistics of DCT coefficient distributed constant.The positive and negative side-play amount span of Off-line data analysis Main Analysis, determines optimized migration weight range, and select optimized migration amount in optimized migration weight range.Research side-play amount and distributed constant and the functional relation with quantization parameter, thus employing least square fitting goes out optimum quantization side-play amount model.
(2) data acquisition and statistical analysis
A) different according to frequency component position, independently adding up respectively DCT coefficient sampling in 4x4 transform block, namely carry out 4x4DCT conversion to a two field picture, the frequency that conversion coefficient is corresponding different, is 16 subbands by conversion coefficient according to frequency partition.Each subband is analyzed separately.Obey the hypothesis of laplacian distribution in DCT coefficient under, obtain the distributed constant of each subband.
B) simulate the behavioural characteristic of SDQ algorithm, Corpus--based Method analytical method maximum just sentencing the double constraints of probability and minimum probability of miscarriage of justice under, estimate optimum dead band side-play amount.Specifically, the quantized result of contrast SDQ and HDQ algorithm, if two amounts result is consistent, adjustment HDQ side-play amount, obtains the offset ranges (δ that HDQ and SDQ quantized result is consistent
min1, δ
max1), this scope bound is determined according to the following formula:
Mod is that remainder operation calculates symbol, and u is DCT coefficient, and q is quantization step, δ
optfor side-play amount.Collect and be allly in scope (δ
min, δ
max) in side-play amount, this scope is called forward migration weight range; If two amounts result is inconsistent, revises the side-play amount of HDQ, force and obtain the quantized result consistent with SDQ algorithm.Offset ranges (the δ revised
min2, δ
max2) bound determines according to the following formula:
This scope is reserve migration weight range.
(3) according to positive opposite phase shift amount (δ
min1, δ
max1) and (δ
min2, δ
max2) determine suitable optimized migration weight range (δ
min, δ
max), namely the common factor of positive and negative offset ranges is optimized migration weight range.Maximum possible span is (0,1), maximum magnitude is divided into N equal portions, to all sample actual range (δ
min, δ
max) contrast, and carry out dividing into groups and classifying.If certain equal portions i belongs to scope (δ in N equal portions
min, δ
max), so this equal portions statistics with histogram value cnt
iadd 1.Histogram analysis process prescription is as follows:
For distributed constant, the various combination of quantization parameter and different predictive modes (intra, inter) obtain separately optimized migration amount segmentation histogram results, and therefrom choosing side-play amount numerical value corresponding to peak value is optimized migration amount.Fig. 2 and 3 is the optimized migration amount statistical value figure of I frame and P/B frame.
(4) under different predictive mode, analyze the functional relation of optimized migration amount and distributed constant and quantization parameter, build the function model of three
.
Wherein Qp>17:
I frame ρ=0.50+0.001* (Qp-17), μ=0.1, β=0.6, γ=0.6, α=128, ξ 1=ξ 2=32, a=35, b=50;
P/B frame ρ=0.48+0.001* (Qp-17), μ=0.1, β=0.6, γ=0.6, α=128, ξ 1=ξ 2=32, a=35, b=50.
Concrete self adaptation side-play amount model value as shown in Figures 4 and 5.
Claims (7)
1. a Video coding hard decision quantization method for content-based self adaptation skew model, comprises the steps:
(1) self adaptation side-play amount modeling;
(2) data acquisition and statistical analysis obtain positive opposite phase shift amount;
(3) according to positive opposite phase shift amount determination optimized migration weight range;
(4) under different predictive mode, analyze the functional relation of optimized migration amount and distributed constant and quantization parameter, build the function model of three.
2. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 1, is characterized in that: the self adaptation side-play amount modeling of described step (1) comprises data statistics, off-line data fractional analysis and heuristic modeling process; Data statistics comprises counting statistics and the rationally candidate offset quantitative statistics of DCT coefficient distributed constant; The positive and negative side-play amount span of Off-line data analysis Main Analysis, determines optimized migration weight range, and select optimized migration amount in optimized migration weight range; Research side-play amount and distributed constant and the functional relation with quantization parameter, thus employing least square fitting goes out optimum quantization side-play amount model.
3. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 1, is characterized in that: the data acquisition of described step (2) and statistical analysis, comprising:
A) different according to frequency component position, independently DCT coefficient in 4x4 transform block sampled and add up respectively, namely 4x4DCT conversion is carried out to a two field picture, the frequency that conversion coefficient is corresponding different, be 16 subbands by conversion coefficient according to frequency partition, each subband is analyzed separately, under obeying the hypothesis of laplacian distribution, obtains the distributed constant of each subband in DCT coefficient;
B) simulate the behavioural characteristic of SDQ algorithm, Corpus--based Method analytical method maximum just sentencing the double constraints of probability and minimum probability of miscarriage of justice under, estimate optimum dead band side-play amount.
4. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 1, it is characterized in that: the behavioural characteristic of step b) simulation SDQ algorithm, be specially the quantized result of contrast SDQ and HDQ algorithm, if two amounts result is consistent, adjustment HDQ side-play amount, obtains the offset ranges (δ that HDQ and SDQ quantized result is consistent
min1, δ
max1), this scope bound is determined according to the following formula:
Mod is that remainder operation calculates symbol, and u is DCT coefficient, and q is quantization step, δ
optfor side-play amount;
Collect and be allly in scope (δ
min, δ
max) in side-play amount, this scope is called forward migration weight range; If two amounts result is inconsistent, revises the side-play amount of HDQ, force and obtain the quantized result consistent with SDQ algorithm; Offset ranges (the δ revised
min2, δ
max2) bound determines according to the following formula:
This scope is reserve migration weight range.
5. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 4, is characterized in that: the common factor of described positive and negative offset ranges is optimized migration weight range.
6. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 5, it is characterized in that: optimized migration weight range maximum possible span is (0,1), maximum magnitude is divided into N equal portions, to all sample actual range (δ
min, δ
max) contrast, and carry out dividing into groups and classifying; If certain equal portions i belongs to scope (δ in N equal portions
min, δ
max), then this equal portions statistics with histogram value cnt
iadd 1; Histogram analysis process prescription is as follows:
For distributed constant, the various combination of quantization parameter and different predictive modes obtain separately optimized migration amount segmentation histogram results, and therefrom choosing side-play amount numerical value corresponding to peak value is optimized migration amount.
7. the Video coding hard decision quantization method of content-based self adaptation skew model as claimed in claim 6, is characterized in that: described function model is;
Wherein Qp>17:
I frame ρ=0.50+0.001* (Qp-17), μ=0.1, β=0.6, γ=0.6, α=128, ξ 1=ξ 2=32, a=35, b=50;
P/B frame ρ=0.48+0.001* (Qp-17), μ=0.1, β=0.6, γ=0.6, α=128, ξ 1=ξ 2=32, a=35, b=50.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510614662.2A CN105208389B (en) | 2015-09-24 | 2015-09-24 | A kind of Video coding hard decision quantization method based on content-adaptive migration model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510614662.2A CN105208389B (en) | 2015-09-24 | 2015-09-24 | A kind of Video coding hard decision quantization method based on content-adaptive migration model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105208389A true CN105208389A (en) | 2015-12-30 |
| CN105208389B CN105208389B (en) | 2018-03-06 |
Family
ID=54955777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510614662.2A Active CN105208389B (en) | 2015-09-24 | 2015-09-24 | A kind of Video coding hard decision quantization method based on content-adaptive migration model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105208389B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106028032A (en) * | 2016-05-24 | 2016-10-12 | 西安电子科技大学 | Coefficient-level adaptive quantization method |
| CN107257461A (en) * | 2017-04-21 | 2017-10-17 | 中国计量大学 | Video encoding hard decision quantization method based on context-adaptive migration model |
| CN107426572A (en) * | 2016-12-30 | 2017-12-01 | 杭州当虹科技有限公司 | A kind of efficiently quick video coding algorithm |
| CN109274967A (en) * | 2018-09-25 | 2019-01-25 | 杭州电子科技大学 | A kind of quick RDOQ quantization zero coefficient ratio estimate method |
| CN111770330A (en) * | 2020-06-10 | 2020-10-13 | 北京达佳互联信息技术有限公司 | Image compression method and device and electronic equipment |
| CN113259667A (en) * | 2021-05-17 | 2021-08-13 | 北京百度网讯科技有限公司 | Video quantization method, video quantization device, electronic equipment and computer-readable storage medium |
| CN113395518A (en) * | 2021-08-13 | 2021-09-14 | 杭州微帧信息科技有限公司 | Method for video coding self-adaptive quantization dead zone |
| CN113422957A (en) * | 2021-05-17 | 2021-09-21 | 北京百度网讯科技有限公司 | Quantization method, apparatus, device and storage medium in video coding |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080226183A1 (en) * | 2007-03-16 | 2008-09-18 | Shawmin Lei | DPCM with Adaptive Range and PCM Escape Mode |
| CN102724495A (en) * | 2012-05-04 | 2012-10-10 | 西安电子科技大学 | Wyner-Ziv frame quantification method based on rate distortion |
| CN103118255A (en) * | 2013-01-25 | 2013-05-22 | 深圳广晟信源技术有限公司 | Self-adaptation quantifying method based on concave model and self-adaptation quantifying device based on concave model |
-
2015
- 2015-09-24 CN CN201510614662.2A patent/CN105208389B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080226183A1 (en) * | 2007-03-16 | 2008-09-18 | Shawmin Lei | DPCM with Adaptive Range and PCM Escape Mode |
| CN102724495A (en) * | 2012-05-04 | 2012-10-10 | 西安电子科技大学 | Wyner-Ziv frame quantification method based on rate distortion |
| CN103118255A (en) * | 2013-01-25 | 2013-05-22 | 深圳广晟信源技术有限公司 | Self-adaptation quantifying method based on concave model and self-adaptation quantifying device based on concave model |
Non-Patent Citations (1)
| Title |
|---|
| 殷海兵，徐宁: "基于自适应预选的快速帧间模式选择算法", 《浙江大学学报》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106028032A (en) * | 2016-05-24 | 2016-10-12 | 西安电子科技大学 | Coefficient-level adaptive quantization method |
| CN106028032B (en) * | 2016-05-24 | 2019-03-26 | 西安电子科技大学 | A kind of coefficient level adaptation quantization method |
| CN107426572A (en) * | 2016-12-30 | 2017-12-01 | 杭州当虹科技有限公司 | A kind of efficiently quick video coding algorithm |
| CN107257461A (en) * | 2017-04-21 | 2017-10-17 | 中国计量大学 | Video encoding hard decision quantization method based on context-adaptive migration model |
| CN107257461B (en) * | 2017-04-21 | 2019-08-20 | 中国计量大学 | Video encoding hard decision quantization method based on context-adaptive migration model |
| CN109274967A (en) * | 2018-09-25 | 2019-01-25 | 杭州电子科技大学 | A kind of quick RDOQ quantization zero coefficient ratio estimate method |
| CN111770330A (en) * | 2020-06-10 | 2020-10-13 | 北京达佳互联信息技术有限公司 | Image compression method and device and electronic equipment |
| CN111770330B (en) * | 2020-06-10 | 2022-11-04 | 北京达佳互联信息技术有限公司 | Image compression method and device and electronic equipment |
| CN113259667A (en) * | 2021-05-17 | 2021-08-13 | 北京百度网讯科技有限公司 | Video quantization method, video quantization device, electronic equipment and computer-readable storage medium |
| CN113422957A (en) * | 2021-05-17 | 2021-09-21 | 北京百度网讯科技有限公司 | Quantization method, apparatus, device and storage medium in video coding |
| CN113422957B (en) * | 2021-05-17 | 2022-11-01 | 北京百度网讯科技有限公司 | Quantization method, apparatus, device and storage medium in video coding |
| CN113395518A (en) * | 2021-08-13 | 2021-09-14 | 杭州微帧信息科技有限公司 | Method for video coding self-adaptive quantization dead zone |
| CN113395518B (en) * | 2021-08-13 | 2022-03-15 | 杭州微帧信息科技有限公司 | Method for video coding self-adaptive quantization dead zone |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105208389B (en) | 2018-03-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105208389A (en) | Video encoding hard decision quantization method based on content self-adaption deviation model | |
| CN102792688B (en) | Data compression for video | |
| CN102835106B (en) | Data compression for video | |
| CN105706450A (en) | Encoder decisions based on results of hash-based block matching | |
| US8630341B2 (en) | Method for training and utilizing separable transforms for video coding | |
| CN104320657B (en) | The predicting mode selecting method of HEVC lossless video encodings and corresponding coding method | |
| JP2001518247A (en) | Digital signal compression encoding method using improved quantization processing | |
| CN1949877A (en) | Adaptive quantization controller and method thereof | |
| CN109286812B (en) | HEVC video quality estimation method | |
| CN103733621A (en) | Rate -distortion optimized video encoding mode selection based on low complexity error propagation tracking | |
| EP3529986A1 (en) | Low complexity mixed domain collaborative in-loop filter for lossy video coding | |
| CN107846593B (en) | Rate distortion optimization method and device | |
| CN110199522A (en) | Processing is used for the reference sample of image block intra prediction | |
| CN104796705A (en) | SSIM (structural similarity) based HEVC (high efficiency video coding) video coding rate distortion optimizing and rate control algorithms | |
| US20220094943A1 (en) | Image processing apparatus, learning appratus, image processing method, learning method and program | |
| CN101977323A (en) | Method for reconstructing distributed video coding based on constraints on temporal-spatial correlation of video | |
| KR100785855B1 (en) | Apparatus and method of embedded quantizaton for the improved SNR scalbilty | |
| CN109076248A (en) | The vector quantization of Video coding is carried out by the code book generated by the training signal selected | |
| Chen et al. | CNN-optimized image compression with uncertainty based resource allocation | |
| KR20110070823A (en) | Method for video codeing and decoding and apparatus for video coding and deconding by using the same | |
| Ma et al. | An adaptive lagrange multiplier determination method for dynamic texture in HEVC | |
| US9503740B2 (en) | System and method for open loop spatial prediction in a video encoder | |
| CN112243123B (en) | HEVC (high efficiency video coding) rate optimization method based on lossless coding | |
| CN104320660B (en) | Rate-distortion optimization method and coding method for lossless video encoding | |
| Jankar et al. | Weighted entropy coding and fractional rate–distortion trade-off for HEVC encoder with emphasis on global bit rate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |