CN105025296A - Advance arithmetic coder and realization method thereof - Google Patents
Advance arithmetic coder and realization method thereof Download PDFInfo
- Publication number
- CN105025296A CN105025296A CN201410183605.9A CN201410183605A CN105025296A CN 105025296 A CN105025296 A CN 105025296A CN 201410183605 A CN201410183605 A CN 201410183605A CN 105025296 A CN105025296 A CN 105025296A
- Authority
- CN
- China
- Prior art keywords
- probabilistic model
- coding
- context
- binaryzation
- module
- 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
Abstract
The invention discloses an advance arithmetic coder and a realization method thereof, belonging to the digit video coding and decoding technical field. The overall structure of the coder employs a parallel efficient design, wherein a three-layer state machine control mechanism employs a three-layer state machine control coding process according to a bit layered structure; the reference data uniform processing technology processes partial syntactic elements into a simpler intermediate form in coding a current macro-block, stores the simpler intermediate form in a Line Buffer to be directly used by a subsequent coding macro-block as a reference without secondary processing, so as to reduce logic circuit costs; the parallel optimization of an algorithm structure employs parallel optimization processing according to a specific algorithm so as to shorten processing time. A probability model access management mechanism is an access control method designed based on the efficient coding characteristic of the advance arithmetic coder, employs two levels of storage modes of RAM and Local Buffer, and shortens the access time of probability model management; the arithmetic coding portion employs a pipelining mode for processing, thereby reducing feedback time and increasing a data processing speed.
Description
Technical field
The present invention relates to digital video decoding technical field, particularly relate to device and its implementation that a kind of advanced mathematical based on AVS encodes.
Background technology
Since nineteen nineties, digital video compaction technique is widely used in the fields such as communication, personal computer, radio and television, consumer electronics, can be rated as the core technology of digital media industry.The source coding standard that current audio frequency and video industry can be selected has four: MPEG-2, MPEG-4, MPEG-4AVC (are called for short AVC, also claim JVT, H.264), AVS, wherein AVS standard is the abbreviation of " information technology advanced audio/video coding " series standard, is the second generation source coding standard that China has independent intellectual property right.AVS is a set of full standard system comprising system, video, audio frequency, media copy management, for digital audio/video industry provides more fully solution.AVS video standard main towards related application such as single-definition/high definition television broadcasting, Web TV and digital storage medias, the computation complexity of cataloged procedure is larger.
Entropy code is a requisite important step in video coding system, mainly by Entropy principle when not losing any information being used for representing that the symbol of element of video sequence changes into compressed bit stream, utilize and remove the object that comentropy redundancy reaches compression.Entropy code mainly comprises huffman coding, adaptive variable length coding and arithmetic coding etc.Wherein, MPEG-2 adopts huffman coding (HuffmanCode), H.264 all adopts with AVS based on context-adaptive variable-length encoding (CAVLC) with based on context adaptive binary arithmetic coding (CABAC).Advanced mathematical coding (AEC) in AVS is that adaptive binary arithmetic coding is designed superior context model and combines with one the method obtained, and its design is based on 3 steps such as binaryzation, context modeling, binary arithmetic codings.AEC is main from two aspects to improve compression ratio: (1) AEC provide non-static statistical model to each syntactic element, encoder can adjust the probabilistic model of syntactic element adaptively according to encoded content and context, adapt to different video contents, the redundant information of syntactic element can be reduced, thus improve compression efficiency.(2) AEC take band as code period, the bin that the syntactic element binarization of one whole piece band obtains is carried out interval iteration division, thus obtain a subinterval, then appoint in this subinterval and get a value and represent this band syntactic element, so, from average, AEC is the numeral of an allocation of symbols non-integer length, overcome the shortcoming that CAVLC coding is necessary for the code word of single allocation of symbols integer code length, more easily approach symbol entropy, obtain higher code efficiency.But the computation complexity of AEC obviously uprises, and proposes challenge for hardware designs.
Summary of the invention
The object of the invention is to carry out considering from chip area and processing speed two aspects provide a kind of advanced mathematical coding structure.
In order to overcome the above problems, the invention discloses a kind of advanced mathematical encoder, primarily of binaryzation and context index computing module, probabilistic model access management module and arithmetic coding module composition.Advanced mathematical coding is that syntax element value is carried out corresponding binaryzation, the corresponding probabilistic model of each binary value (bin) obtained, the probabilistic model obtained and binary value are carried out arithmetic coding, obtains final code stream (bit).According to the difference of functions of modules, advanced mathematical encoder adopts parallel efficient design mode.Because the binarization method of different syntactic element is different with the method for probabilistic model index selection, its determination methods is that frame field type, the macro block (mb) type of current coding macro block and the syntax element type etc. of present encoding according to band decides.So only syntactic element is carried out binaryzation in design, and based on context select the probabilistic model index of each bin.To obtain after binary value and corresponding probabilistic model index, without the need to seeing condition again, can directly encoding.So choosing of probabilistic model is carried out outlet process with arithmetic coding unification.
Further, advanced mathematical encoder adopts state control mode to encode to syntactic element, and this module needs to know that some specific information judge the syntactic element that will encode.For AVS standard, advanced mathematical coding is in units of band, band is made up of several macro blocks, and each macro block contains different syntactic elements, in advanced mathematical coding, the coded sequence of syntactic element is fixing, but according to the difference of frame field type and macro block (mb) type, the syntactic element comprised in macro block is different.So adopt three layer state machines to carry out control coding order in advanced mathematical coding control module.Ground floor is band redirect, and the second layer is the redirect of a macro block, and third layer is the redirect of syntax element encodes order.
Based on the context coding feature of arithmetic coding, when utilizing line buffer storing one row macro block, be to restore after data processing in line buffer.Like this when encoding current macroblock needs its top macro block, data can be taken out from line buffer directly to use, without the need to carrying out a series of Logic judgment analysis again, the mode stored again after such united analysis data reduces logical complexity when memory space is suitable, reduces logical circuit area.
Binaryzation and context index computing module inside adopt parallel data processing, reduce operation time, ensure processing speed.This module will be averaging one-period (cycle) and export a bin and corresponding probabilistic model index ctxIdx, parallel data processing is made according to the characteristic of syntactic element binaryzation when design module hardware configuration, can the more data of parallel processing within the identical time, ensure design module performance.
Probabilistic model administration module, first will choose probability according to probabilistic model index and encode.Deposit in the RAM of probabilistic model the probability of renewal being deposited to get back to after probability updating.In order to reduce the feedback processing time, adopting RAM and local cache secondary storage mode and increasing a probabilistic model control module to manage when get probabilistic model, when depositing probabilistic model.Its major function is, probabilistic model as used with last bin in current probability model is identical, then need not get new probability, directly upgrade.When running into different probability model, get new probabilistic model, while carrying out probability updating, the probabilistic model after last renewal is deposited back in RAM, utilize parallel processing and control mode to reach the object reducing the processing time.
In advanced mathematical coding, arithmetic coding module (BAC) is the core of algorithm, due to the advanced mathematical coding not look-up table in AVS standard, but will through calculating update probability, computation complexity uprises relatively.So arithmetic coding module adopts the design of streamline.Overall processing speed is improved.
Accompanying drawing explanation
When considered in conjunction with the accompanying drawings, by referring to detailed description below, more completely can understand the present invention better and easily learn wherein many adjoint advantages, but accompanying drawing described herein is used to provide a further understanding of the present invention, form a part of the present invention, schematic description and description of the present invention, for explaining the present invention, does not form inappropriate limitation of the present invention, wherein:
Fig. 1, advanced mathematical encoder basic structure schematic diagram;
Fig. 2, advanced mathematical encoder concrete structure schematic diagram;
Fig. 3, the storage schematic diagram of macro block adjacent block;
Storage mode schematic diagram in Fig. 4, line buffer;
Fig. 5, advanced mathematical coding module ground floor state machine diagram;
Fig. 6, advanced mathematical coding module second layer state machine diagram;
Fig. 7, advanced mathematical coding module third layer view;
Fig. 8, binaryzation and context index computing module schematic diagram;
Fig. 9, probabilistic model access management module diagram;
Figure 10, arithmetic coding module schematic diagram.
Embodiment
Be described referring to Fig. 1-10 pairs of embodiments of the invention.
For enabling above-mentioned purpose, feature and advantage become apparent more, and below in conjunction with the drawings and specific embodiments, the present invention is further detailed explanation.
As shown in Figure 1, a kind of realization of the advanced mathematical encoder based on AVS that the present invention proposes, mainly comprises three layer state machine controlling mechanisms, reference data unifies treatment technology, algorithm structure parallel optimization and arithmetic coding efficient stream water-bound five kinds of technology.Just implementation of the present invention does example elaboration respectively below.Coding-control as shown in Figure 1 for the input of binaryzation and context modeling have employed three layer state machine controlling mechanisms, binaryzation and context modeling have employed reference data and unify treatment technology and algorithm structure parallel optimization, specifically comprise the participation analyzing syntactic element, binaryzation and context index computing module and line buffer.Probabilistic model access management have employed probabilistic model access management mechanism, specifically comprises the participation of analysis context index module, context RAM1 and context RAM2 and local cache (Local Buffer).Arithmetic coding have employed efficient stream water-bound.
The operation principle of encoder shown in Fig. 1 is: the syntax element value of input is carried out binaryzation and produce binary value, and based on context select the probabilistic model index corresponding to binary value, according to the corresponding probabilistic model of probabilistic model index selection, the probabilistic model of binary value and its correspondence is carried out arithmetic coding, and the code stream after output encoder; After completing described arithmetic coding, also comprise the probabilistic model corresponding to described binary value upgrade.
A kind of advanced mathematical encoder as shown in Figure 2, primarily of binaryzation and context index computing module, probabilistic model access management module and arithmetic coding module composition, described binaryzation and context index computing module are used for syntax element value being carried out binaryzation and produce binary value, and based on context select the probabilistic model index corresponding to binary value; Described probabilistic model access management module, for according to the corresponding probabilistic model of probabilistic model index selection, upgrades for the probabilistic model corresponding to described binary value; Described arithmetic coding module is used for the probabilistic model of binary value and its correspondence to carry out arithmetic coding, and the code stream after output encoder; Described encoder using binary value and probabilistic model index FIFO as data buffer, binaryzation and context index computing module, probabilistic model access management module and arithmetic coding module concurrent operation.Described binaryzation and context index computing module inside adopt binaryzation and based on context calculating probability model index parallel processing, in the inside of binaryzation, the binaryzation of macroblock coding template can directly obtain, the binaryzation of motion vector components difference (mvd) need by calculating, so by the binaryzation of cbp and mvd and top block and left side blocks of data parallel processing can be organized, in the guaranteed situation of hardware resource area, reduce operation time, ensure processing speed.
The overall structure of Fig. 2 arithmetic encoder, arithmetic encoder adopts coding control module to control whole cataloged procedure.First arithmetic encoder sends request as dynamic random access memory, all data that a macro block will be encoded can be delivered in syntactic element data buffer by dynamic random access memory, carry out control analysis syntactic element module by coding control module from buffer, read data and analyze, use top block will be read stored in line buffer in order to next code after unified for data process.Binaryzation and context index computing module can read the information of current syntax element and the syntax element information of top block subsequently, carry out binaryzation, and the probabilistic model index of each binary value is calculated according to top block and left side block, binary value and probabilistic model index are spliced, exports to binary value and probabilistic model index FIFO cushions.Analysis context index module reads each binary value of data post analysis and corresponding probabilistic model index from buffer, deposit the RAM (context RAM1 and context RAM2) of probabilistic model from two and extract after corresponding probabilistic model, export to binary arithmetic coding module to encode, being deposited by the code stream spued after coding gets back in dynamic memory, obtains last coded file.In Fig. 2, need to learn concrete syntax element type in part A, frame, field type and macro block (mb) type etc. carry out binaryzation to syntactic element and context index calculates.But in B and C part just without the need to knowing the occurrence of these variablees again, coding method unitizes, as long as learn each bin (binary value after binaryzation) and corresponding probabilistic model index address, just can encode.Feature according to above arithmetic coding carries out function division, adopts parallel efficient design structure.Wherein binary value and probabilistic model index FIFO play cushioning effect, and two different modules of processing speed can better be coordinated.As long as part A sees FIFO, discontented just can continuation processes syntactic element, by the bin after binaryzation and corresponding probabilistic model index press-in FIFO, as long as and coding below to detect FIFO not empty, just can read also coded data continuously.Such design can make internal structure relatively independent, and judges whether to need work by best detection signal.
Because encoding current macroblock carrys out select probability model according to its top macro block and left side macro block information, so need top macro block and the access of left side macro block.So as shown in Figure 3, left side macro block adopts buffer to store, and top macro block adopts a line buffer to store.When storing data, data first store by again, like this when next code, can take out data and directly use from line buffer, without the need to carrying out a series of Logic judgment analysis again, as shown in Figure 4, we are divided into four fritters, according to the difference of macro block (mb) type a macro block, corresponding syntactic element is stored to four fritters, the macro block (mb) type of such as 8X16, No. 0 block is identical with the storage data of No. 2 blocks, and No. 1 block is identical with the storage data of No. 3 blocks.Only need store the information of No. 2 blocks and No. 3 blocks when storing top macro block, and left side macro block stores No. 1 block and No. 3 blocks.The mode stored again after such united analysis data reduces logical complexity when memory space is suitable, reduces circuit area overhead.
According to arithmetic coding with band (slice) for unit, slice comprises macro block, comprises the characteristic of syntactic element in macro block, controls with three grades of state machines the coding of advanced mathematical coding module.As shown in Figure 5, ground floor is slice state transition, starts a macro block of encoding after entering AEC MB state, in this condition, separates the state machine of the second layer for a macroblock coding.Fig. 6 is the second layer is macroblock status redirect, under MB One state, separates third layer state machine.Fig. 7 third layer is then the state transition of syntactic element, mainly carries out order redirect to the syntactic element in a macro block according to condition.
As shown in Figure 8, this module adopts binaryzation and probabilistic model index control module to control whole process, first reads linebuffer, obtains the data of top block for binaryzation and probabilistic model index calculation module.Carry out binaryzation by binaryzation advanced processing module in advance to some special syntactic elements while unifying process by the syntax element value of organising data module to current macro and top macro block subsequently, the data after process export to binaryzation and probabilistic model index module.Control module is according to the number of state transition Driving technique module according to current syntactic element Data-Statistics binary value, the data of having added up are given binaryzation and context index output module, the statistics number of the contextual information sum counter that this module provides according to organising data, carries out corresponding binary value to syntactic element and probabilistic model index exports.A bin and corresponding probabilistic model index (ctxIdx) is exported because this module will be averaging one-period (cycle), make parallel data processing when design module hardware configuration according to the characteristic of syntactic element binaryzation, reduce data processing time.As use the syntactic element of unitary code binaryzation method, first count with counting module, then export corresponding bin.But the binarization method of mvd (motion vector difference) is more special, if the value of synElVal (syntactic element) is more than or equal to 3 and the value of synElVal is odd number, first four of binary symbols string is " 1110 ", and subsequent bit is 0 rank Exp-Golomb of (synElVal-3)/2 correspondence; If the value of synElVal is greater than 3 and the value of synElVal is even number, first four of binary symbols string is " 1111 ", and subsequent bit is 0 rank Exp-Golomb of (synElVal-3)/2 correspondence.So can Columbus's binarizing portion advanced processing, and organising data executed in parallel.The maximum of syntactic element CBP (macroblock coding template) is 63, and its binarization method is fixed, and does not have rule, so also can be placed on parallel processing above.Such realization can reduce the processing time, ensures efficient processing performance.
After extraction probabilistic model carries out arithmetic coding probability updating, need to upgrade in time to probability, result in whole processing speed and reduce.In order to address this problem, as shown in Figure 9, the present invention adopts two RAM and local caches secondary storage structure, and adds the probabilistic model access management module controls access time.Wherein context RAM1 and context RAM2 is two block RAMs, is used for storage 195 probabilistic models respectively.Probabilistic model access management module is control module, in order to avoid by probability with the probabilistic model selecting repetition after newly going back again, after receiving probability index, this module can judge, if use same probabilistic model, direct with the probability after previous renewal, need not read from RAM, if probabilistic model used is not same, then in RAM, choose corresponding probabilistic model.
Arithmetic coding is algorithm nucleus module, and as shown in Figure 10, described arithmetic coding module adopts the method for designing of Pyatyi flowing water, and first order flowing water is used for probabilistic model to upgrade; Module in the flowing water of the second level is in order to avoid selecting the probabilistic model of repetition after being gone back by probability updating again, after receiving probabilistic model index, can judge, if use same probabilistic model, direct with the probabilistic model after previous renewal, need not read from context RAM, if probabilistic model used is not same, then in context RAM, choose corresponding probabilistic model; Third level flowing water mainly upgrades region, first calculates region, gives area update module safeguard the data after calculating region, and the data after maintenance can feed back to area calculation module and export; Fourth stage flowing water upgrades region starting point; Level V flowing water spues after mainly completing coding reformation normalization code stream.Probability updating adopts non-multiplication but carries out addition and subtraction after changing into log-domain to calculate new probability, though simplify, but algorithm is still very complicated.But when we find update probability, a lot of computing is not fed back, and is single renewal budget, so the present invention adopts flowing water method, by complex calculations unification.After binaryzation above, by the bin after binaryzation and probabilistic model index stored in FIFO, the processing speed of this process can be very fast.And arithmetic coding adopts flowing water design below, the bin that can constantly encode in FIFO, is improved like this in bulk velocity.
Last it is noted that obviously, above-described embodiment is only for example of the present invention is clearly described, and the restriction not to execution mode.For those of ordinary skill in the field, can also make other changes in different forms on the basis of the above description.Here exhaustive without the need to also giving all execution modes.And thus the apparent change of amplifying out or variation be still among protection scope of the present invention.
Claims (7)
1. an advanced mathematical encoder, it is characterized in that: primarily of binaryzation and context index computing module, probabilistic model access management module and arithmetic coding module composition, described binaryzation and context index computing module are used for syntax element value being carried out binaryzation and produce binary value, and based on context select the probabilistic model index corresponding to binary value; Described probabilistic model access management module, for according to the corresponding probabilistic model of probabilistic model index selection, upgrades for the probabilistic model corresponding to described binary value; Described arithmetic coding module is used for the probabilistic model of binary value and its correspondence to carry out arithmetic coding, and the code stream after output encoder;
Described encoder using binary value and probabilistic model index buffer as data buffer, binaryzation and context index computing module, probabilistic model access management module and arithmetic coding module concurrent operation.
2. advanced mathematical encoder according to claim 1, it is characterized in that: described binaryzation and context index computing module inside adopt binaryzation and based on context calculating probability model index parallel processing, the concrete inside in binaryzation, by the binaryzation of macroblock coding module and motion vector components difference with organize top block and left side blocks of data parallel processing.
3. advanced mathematical encoder according to claim 1, it is characterized in that: adopt LineBuffer to store upper adjacent block, reference data is utilized to unify treatment technology, restore after data processing is carried out in unification Line Buffer, concrete grammar is that a macro block is divided into four fritters 0,1,2,3, the information of four fritters is determined according to the type of macro block, storage only needs storage No. 2 blocks and No. 3 blocks during adjacent macroblocks, only need store No. 1 block and No. 3 blocks when storing left adjacent macroblocks.
4. advanced mathematical encoder according to claim 1, it is characterized in that: described probabilistic model access management module adopts two-level memory, comprise context RAM and local caches, the wherein context probability modeling used for storing nearest meeting of local caches, context RAM is used for depositing 195 context probability modeling information, decide when to extract probabilistic model with probabilistic model access control module between context RAM and local caches, when the probabilistic model after renewal is deposited back.
5. advanced mathematical encoder according to claim 1, is characterized in that: described arithmetic coding module adopts the method for designing of Pyatyi flowing water, and first order flowing water is used for probabilistic model to upgrade; Module in the flowing water of the second level is in order to avoid selecting the probabilistic model of repetition after being gone back by probability updating again, after receiving probabilistic model index, can judge, if use same probabilistic model, direct with the probabilistic model after previous renewal, need not read from context RAM, if probabilistic model used is not same, then in context RAM, choose corresponding probabilistic model; Third level flowing water mainly upgrades region; Fourth stage flowing water upgrades region starting point; Level V flowing water spues after mainly completing coding reformation normalization code stream.
6. the implementation method of an advanced mathematical encoder, it is characterized in that: the syntax element value of input is carried out binaryzation and produce binary value, and based on context select the probabilistic model index corresponding to binary value, according to the corresponding probabilistic model of probabilistic model index selection, the probabilistic model of binary value and its correspondence is carried out arithmetic coding, and the code stream after output encoder; After completing described arithmetic coding, also comprise the probabilistic model corresponding to described binary value upgrade.
7. method according to claim 6, it is characterized in that: described binaryzation of the syntax element value of input being carried out produces binary value, adopt three layer state machines to control, the coding of ground floor state machine major control band, the initialization and the band end-of-encode needs that comprise band flush; The coding of second layer state machine major control macro block, whether comprise macro block is last macro block, if last macro block wants direct coding 1, otherwise coding 0; The order of the syntactic element in third layer state machine major control macro block, decides type and the order of syntax elements encoded according to frame/field type, macro block (mb) type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410183605.9A CN105025296B (en) | 2014-04-30 | 2014-04-30 | A kind of advanced mathematical encoder and its implementation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410183605.9A CN105025296B (en) | 2014-04-30 | 2014-04-30 | A kind of advanced mathematical encoder and its implementation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105025296A true CN105025296A (en) | 2015-11-04 |
| CN105025296B CN105025296B (en) | 2018-02-23 |
Family
ID=54414957
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410183605.9A Active CN105025296B (en) | 2014-04-30 | 2014-04-30 | A kind of advanced mathematical encoder and its implementation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105025296B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107071494A (en) * | 2017-05-09 | 2017-08-18 | 珠海市杰理科技股份有限公司 | The generation method and system of the binary syntax element of video frame image |
| CN107659814A (en) * | 2017-09-21 | 2018-02-02 | 深圳市德赛微电子技术有限公司 | Entropy decoding structure in a kind of bimodulus decoder of AVS and MPEG 2 |
| CN111683253A (en) * | 2020-06-12 | 2020-09-18 | 浪潮(北京)电子信息产业有限公司 | Parameter set decoding method and device |
| CN111787325A (en) * | 2020-07-03 | 2020-10-16 | 北京博雅慧视智能技术研究院有限公司 | Entropy encoder and encoding method thereof |
| WO2023197104A1 (en) * | 2022-04-11 | 2023-10-19 | Oppo广东移动通信有限公司 | Coding method, decoding method, encoders, decoder and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231830A (en) * | 2010-11-23 | 2011-11-02 | 浙江大学 | Arithmetic unit used for context arithmetic encoding and decoding |
| US20120293343A1 (en) * | 2011-05-17 | 2012-11-22 | National Cheng Kung University | Method for high throughput parallel avc/h.264 context-based adaptive binary arithmetic coding decoder |
| CN103227924A (en) * | 2013-05-07 | 2013-07-31 | 北京大学 | Arithmetic coder and coding method |
-
2014
- 2014-04-30 CN CN201410183605.9A patent/CN105025296B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231830A (en) * | 2010-11-23 | 2011-11-02 | 浙江大学 | Arithmetic unit used for context arithmetic encoding and decoding |
| US20120293343A1 (en) * | 2011-05-17 | 2012-11-22 | National Cheng Kung University | Method for high throughput parallel avc/h.264 context-based adaptive binary arithmetic coding decoder |
| CN103227924A (en) * | 2013-05-07 | 2013-07-31 | 北京大学 | Arithmetic coder and coding method |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107071494A (en) * | 2017-05-09 | 2017-08-18 | 珠海市杰理科技股份有限公司 | The generation method and system of the binary syntax element of video frame image |
| CN107071494B (en) * | 2017-05-09 | 2019-10-11 | 珠海市杰理科技股份有限公司 | The generation method and system of the binary syntax element of video image frame |
| CN107659814A (en) * | 2017-09-21 | 2018-02-02 | 深圳市德赛微电子技术有限公司 | Entropy decoding structure in a kind of bimodulus decoder of AVS and MPEG 2 |
| CN111683253A (en) * | 2020-06-12 | 2020-09-18 | 浪潮(北京)电子信息产业有限公司 | Parameter set decoding method and device |
| CN111787325A (en) * | 2020-07-03 | 2020-10-16 | 北京博雅慧视智能技术研究院有限公司 | Entropy encoder and encoding method thereof |
| CN111787325B (en) * | 2020-07-03 | 2022-03-08 | 北京博雅慧视智能技术研究院有限公司 | Entropy encoder and encoding method thereof |
| WO2023197104A1 (en) * | 2022-04-11 | 2023-10-19 | Oppo广东移动通信有限公司 | Coding method, decoding method, encoders, decoder and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105025296B (en) | 2018-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103733622B (en) | Context initialization in entropy code | |
| US20180176556A1 (en) | Low-complexity sign prediction for video coding | |
| CN104221377B (en) | A kind of video decoding method, device and computer-readable storage medium | |
| CN105025296A (en) | Advance arithmetic coder and realization method thereof | |
| KR101941955B1 (en) | Recursive block partitioning | |
| CN100547615C (en) | Become the encoding scheme of the data stream of graphical model during expression | |
| JP6130839B2 (en) | Entropy coding | |
| CN101115208B (en) | Picture coding apparatus | |
| CN105357540A (en) | Method and apparatus for decoding video | |
| CN101427583B (en) | Arithmetic decoding method and device | |
| US10791341B2 (en) | Binary arithmetic coding with progressive modification of adaptation parameters | |
| CN103227924B (en) | A kind of arithmetic encoder and coded method | |
| CN104365099A (en) | Parameter update method for entropy coding and decoding of conversion coefficient level, and entropy coding device and entropy decoding device of conversion coefficient level using same | |
| CN102238387A (en) | Video entropy code as well as entropy coding method, device and medium | |
| CN104539949B (en) | The method and device of quick partitioning based on edge direction in HEVC screen codings | |
| CN106576168A (en) | Image compression/decompression device | |
| CN102238383B (en) | For the multibus system architecture of Video Codec | |
| CN101179729A (en) | Interframe mode statistical classification based H.264 macroblock mode selecting method | |
| WO2024037091A1 (en) | Point cloud processing method and apparatus, computer device, and storage medium | |
| CN108353174A (en) | Parallel arithmetic decoding technique | |
| CN106063269A (en) | Moving image coding device, moving image decoding device, moving image coding method, and moving image decoding method | |
| Zhao et al. | An optimization method of WebP images lossy compression algorithm for FPGAs | |
| CN112073729B (en) | Model updating method, device, electronic equipment and computer readable storage medium | |
| Deprá et al. | A novel hardware architecture design for binary arithmetic decoder engines based on bitstream flow analysis | |
| Cui et al. | An efficient zigzag scanning and entropy coding architecture design |
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 |