CN105007486A - Method of coding and decoding images, coding and decoding device and computer programs - Google Patents
Method of coding and decoding images, coding and decoding device and computer programs Download PDFInfo
- Publication number
- CN105007486A CN105007486A CN201510340428.5A CN201510340428A CN105007486A CN 105007486 A CN105007486 A CN 105007486A CN 201510340428 A CN201510340428 A CN 201510340428A CN 105007486 A CN105007486 A CN 105007486A
- Authority
- CN
- China
- Prior art keywords
- group
- probability data
- piece
- block
- decoding
- 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
A method of coding and decoding images, a coding and decoding device and computer programs are disclosed. The invention relates to a method of coding at least one image. The method comprises the steps of splitting the image into a plurality of blocks, grouping said blocks into a predetermined number of subsets of blocks, coding each of said subsets of blocks in parallel, the blocks of a subset considered being coded according to a predetermined sequential order of traversal. The coding step comprises, for a current block of a subset considered, the sub-step of predictive coding of said current block with respect to at least one previously coded and decoded block, and the sub-step of entropy coding of said current block on the basis of at least one probability of appearance of a symbol. In the case where the current block is the first block to be coded of a subset considered, said probability of appearance of a symbol is that which was calculated for a predetermined coded and decoded block of at least one other subset. In the case where the current block is a block of said subset considered, other than said first block of the latter, said probability of appearance of a symbol is that which was calculated for at least one coded and decoded block belonging to said same subset.
Description
The application is application number is 201280021355.8, and the applying date is on February 23rd, 2012, is entitled as the divisional application of the Chinese invention patent application of " method of Code And Decode image, Code And Decode equipment, computer program ".
Technical field
The present invention generally belongs to the field of image procossing, belongs to the Code And Decode of digital picture and digital image sequence more accurately.
The present invention especially can be used to the Video coding realized in current video encoder (MPEG, H.264 etc.) or the video encoder (ITU-T/VCEG (H.265) or ISO/MPEG (HVC)) in future thus.
Background technology
Current video encoder (MPEG, H264 etc.) uses the block (block-wise) of pressing of video sequence to represent.Image is split as macro block, and each macro block self is split as block, and each piece or macro block by image or inter picture prediction encode.So, specific image is encoded by spatial prediction (infra-frame prediction), and other images motion compensation well known by persons skilled in the art auxiliary under encoded by time prediction (inter prediction) about one or more coding-decoded reference pictures.In addition, for each piece, can encode to residual block, this residual block and original block deduct predict corresponding.After optional conversion, the coefficient of this block can be quantized, and is then coded by entropy device coding.
Encoded available with specific piece of decoding before infra-frame prediction and inter prediction need, thus be used to predict current block in decoder or encoder.Figure 1A represents the schematic example of such predictive coding, wherein, and image I
nbe divided into block, the current block MB of this image
istand three block MBr with the predetermined quantity of Code And Decode before such as represented by hatched arrows
1, MBr
2, MBr
3relevant predictive coding.Above-mentioned three blocks are particularly including being disposed immediately in current block MB
ithe block MBr on the left side
1, and be disposed immediately in current block MB respectively
itop and top-right two block MBr
2and MBr
3.
Here more special it is of concern that entropy coder.Entropy coder comes information coding according to the order arrived.Typically realize the traversal line by line of the block of " raster scan " type, if Figure 1A is by with reference to shown in (label) PRS, from the BOB(beginning of block) in the upper left corner of image.For each piece, represent that each item of information (block type, predictive mode, residual coefficients etc.) needed for this block is sequentially distributed to entropy coder.
Known a kind of enough complicated significance arithmetic encoder, be called as " CABAC " (" context adaptation binary arithmetic coder "), it is introduced in AVC compression standard (be also referred to as ISO-MPEG4 the 10th part and ITU-T H.264).
Entropy coder realizes various design:
-arithmetic coding: such as initial at document J.Rissanen and G.G.Langdon Jr, " Universalmodeling and coding; " IEEE Trans.Inform.Theory, vol.IT-27, the encoder described in pp.12 – 23, Jan.1981 uses the probability of occurrence of this symbol to come encoding symbols;
-context adapts to: this comprises the probability of occurrence adapting to the symbol that will encode.On the one hand, realize learning fast.On the other hand, according to the state of the information of encoding before, specific context is used to coding.There is corresponding intrinsic symbol appearing probability in each context.Such as, context corresponds to the type (expression of residual coefficient, the signaling of coding mode etc.) of coded identification according to given configuration, or corresponds to the state (such as from the quantity etc. of " in frame " pattern of neighbor choice) of neighbours;
-binarization: the symbol that encode is converted into the form of Bit String.Subsequently, this each bit is distributed to binary system entropy coder in succession.
So this entropy coder realizes a kind of system, for the probability that the symbol learned fast to encode for the context considered before is relevant for each context used.This learns the coded sequence based on these symbols.Typically, traversing graph picture is carried out according to the order of " raster scan " type as mentioned above.
When encoding to the given symbol b that can equal 0 or 1, in the following manner for current block MB
iupgrade the probability of occurrence P of this symbol
ilearn:
Wherein, α is predetermined value, such as 0.95, and P
i-1it is the symbol appearing probability calculated when this symbol finally occurs.
Figure 1A represents the schematic example of such entropy code, wherein, and image I
ncurrent block MB
ibe coded by entropy.At current block MB
ientropy code when starting, the symbol appearing probability used is the symbol appearing probability obtained after the coding of the block of Code And Decode before, according to the traversal line by line of the block of above-mentioned " raster scan " type, current block MB should be adjacent at the block of Code And Decode before
ibefore.Only clear in order to illustrate, represent learning based on block-by-block is dependent for specific block by thin-line arrow in Figure 1A.
The defect of the entropy code of the type is the following fact, to be positioned at row start encoding symbols time, the probability of use, mainly corresponding to the viewed probability of symbol at the end being positioned at previous row, travels through relevant to " raster scan " of block.Now, because the possible space of symbol probability changes (such as, for the symbol relevant to movable information item, the motion being positioned at image right part from observe in left part can be different, and thus be similar for consequent local probability), can observe the disappearance of the local adaptation of probability, this has when encoding the risk causing validity to lose.
At internet address
http:// wftp3.itu.int/av-arch/jctvc-site/2010_04_A_Dresden/JCTVC-A114-Annex a.docdescribe a kind of coding method in the document " Annex A:CDCM Video Codec DecoderSpecification " that (on February 8th, 2011) provides, it alleviates above-mentioned defect.As shown in Figure 1B, the coding method described in above document comprises:
-by image I
ncutting is the step of multiple pieces,
Three block MBr of the predetermined quantity of Code And Decode before that-combination is such as represented by hatched arrows
1, MBr
2, MBr
3to the current block MB of this image
icarry out the step of predictive coding.Above-mentioned three blocks are particularly including being disposed immediately in current block MB
ithe block MBr on the left side
1, and be disposed immediately in current block MB respectively
itop and top-right two block MBr
2and MBr
3,
-to image I
nblock carry out the step of entropy code, accordingly, when these blocks are available, every block uses respectively for the symbol appearing probability that block and the block that is encoded and decoding in left side that is disposed immediately in current block of the encoded and decoding be disposed immediately in above current block calculate.In order to make the latter more clear, carry out by thin-line arrow the use partly representing this symbol appearing probability in fig. ib.
The benefit of this entropy code is, which utilizes the probability produced by the immediate environment of current block, makes thus to obtain higher coding efficiency.In addition, make may to the subset parallel encoding of the paired adjacent block of predetermined quantity for the coding techniques of use.In the example that Figure 1B shows, three subset SE1, SE2 and SE3 are by parallel encoding, and a line block that each subset is illustrated by the broken lines in this example embodiment is formed.Certainly, this coding need to lay respectively at the top of current block and top-right piece available.
The defect of this parallel encoding technology is, in order to the symbol appearing probability allowing access needle to calculate the block be disposed immediately in above current block, needs to store some probability associated with a line block.If such as consider the block of the second row SE2 in fig. ib, by using first piece of symbol appearing probability calculated for the first row SE1 before, first piece of this row is coded by entropy.When completing the coding of first piece of the second row, the state of the value V1 of probability of occurrence is stored in buffer storage MT.Subsequently, by using simultaneously for second piece of the first row SE1 and first piece of symbol appearing probability calculated of the second row SE2, second piece of the second row SE2 is coded by entropy.When completing the coding of second piece of the second row, the state of the value V2 of probability of occurrence is stored in buffer storage MT.This process lasts, until last block of the second row SE.Due to the quantity of probability very large (there is the as many probability of combination with the quantity of syntactic element and the quantity of associated context), these probability of whole a line to be stored in memory resource aspect very expensive.
Summary of the invention
A target of the present invention is the defect making up above-mentioned prior art.
For this reason, theme of the present invention relates to a kind of method of encoding at least one image, comprises the following steps:
-image is divided into multiple pieces,
-described piece is grouped into the subset of the block of predetermined quantity,
-to each parallel encoding in the subset of described piece, the block in the subset considered is encoded according to the predefined procedure of traversal, and for the current block in considered subset, described coding step comprises substep:
About the block of Code And Decode before at least one, predictive coding is carried out to described current block,
By the symbol appearing probability before learning at least one, entropy code is carried out to current block,
The method according to the present invention part that merits attention is:
When-will encode in the subset that current block is considered first piece, symbol appearing probability is the symbol appearing probability calculated for the predetermined block of the Code And Decode at least one other subset,
When-block in the subset that current block is considered except first piece, symbol appearing probability is the symbol appearing probability calculated at least one block that is encoded and decoding belonging to same subsets.
This layout makes the symbol appearing probability that may store lesser amt in the buffer storage of encoder, because the entropy code of the current block except first piece of subset of block no longer must need to use the symbol appearing probability calculated about block that is encoded and decoding before, should before block that is encoded and decoding be arranged in another subset and be positioned at above current block.
This layout also makes to keep existing compression performance, because the entropy code of current block uses the symbol appearing probability calculated for the block of Code And Decode before another in the subset belonging to current block, and finally realize learning by update probability, thus the latter meets the statistics of vision signal.
The key benefit of first piece of symbol appearing probability calculated for other subsets described is used to be during the first current block entropy code in the subset of considered block, save this buffer storage by only storing more newly arriving of described symbol appearing probability in the buffer storage of encoder, and the symbol appearing probability that other continuous blocks of other subsets described are learned need not be considered.
During the entropy code of the first current block of the subset of considered block, use the key benefit of the symbol appearing probability calculated for the block (such as second piece) in other subsets described except first piece to be, obtain to symbol appearing probability more accurately and therefore better learn, bring better video compression performance thus.
In certain embodiments, the described Code And Decode block belonging to same subsets with the current block that will encode except first piece in described subset is the nearest neighbor block of the current block that will encode.
This layout makes may only store the symbol appearing probability learned during the entropy code of first piece in considered subset thus, because under this special case, only consider for being positioned at above the first current block and the symbol appearing probability that calculates of the block belonging to another subset.The optimization that the size that this results in the memory resource of encoder reduces.
In another specific embodiments, the predictive coding of the block in considered subset is by when before the predetermined quantity beyond the subset about described consideration, the block of Code And Decode performs, and the parallel encoding of the block in the subset of described consideration is so that relative to the subset being adjacent to block above under the order performing parallel encoding, the block that offsets predetermined quantity performs.
This layout allows the subset for the current block that will encode to realize the synchronous of the processing procedure of the block in the subset of the block under the order performing parallel encoding before current subnet, makes the availability that may ensure one or more pieces in the previous subset of the coding being used to current block thus.Pass through which, advantageously can be ignored in the step (step realized in such as the parallel encoder in prior art) of the availability verifying this block in subset above, allowed the acceleration in the processing time needed for block being processed in encoder according to the present invention thus.
Relatively, the invention still further relates to a kind of equipment that at least one image is encoded, comprising:
-parts for image being divided into multiple pieces,
-for block being grouped into the parts of the subset of the block of predetermined quantity,
-parts to each parallel encoding in the subset of block, the block in the subset considered is encoded according to the predefined procedure of traversal, and for the current block in considered subset, described addressable part comprises:
About the block of Code And Decode before at least one, current block is carried out to the subassembly of predictive coding,
Based at least one symbol appearing probability, current block is carried out to the subassembly of entropy code,
The described encoding device part that merits attention is:
When-will encode in the subset that current block is considered first piece, for the entropy code of the first current block, the subassembly of entropy code considers the symbol appearing probability calculated for the predetermined block of the Code And Decode at least one other subset,
When-block in the subset that current block is considered except first piece, for the entropy code of current block, the subassembly of entropy code considers the symbol appearing probability calculated at least one the Code And Decode block belonging to same subsets.
By corresponding mode, the invention still further relates to a kind of to representing the method that the stream of at least one coded image is decoded, comprising the following steps:
-identify the subset of the block of predetermined quantity that will decode in the picture,
-to decode to the part parallel of each stream associated in the subset of block, the block in the subset considered is decoded according to the predefined procedure of traversal, and for the current block in considered subset, described decoding step comprises substep:
Based at least one symbol appearing probability, entropy decoding is carried out to current block,
Block about decoding before at least one carries out prediction decoding to current block,
This coding/decoding method part that merits attention is:
When-will decode in the subset that current block is considered first piece, symbol appearing probability calculates for the predetermined block of decoding of at least one other subset,
When-block in the subset that current block is considered except in this subset first piece, symbol appearing probability calculates for block of at least one decoding belonging to same subsets.
In certain embodiments, with the nearest neighbor block that the decoding block wanting decoded current block to belong to identical subset beyond in described subset first piece is the current block that will decode.
In another specific embodiment, the prediction decoding of the block in considered subset is wanted for when before the predetermined quantity except considered subset, the block of Code And Decode performs, and the parallel decoding of the block in the subset considered is that the block that offsets predetermined quantity performs relative to the subset being adjacent to block above under the order performing parallel decoding.
Relatively, the invention still further relates to a kind of to representing the equipment that the stream of at least one coded image is decoded, comprising:
-identify the parts of the subset of the block of predetermined quantity that will decode in the picture,
-to the parts of decoding with the part parallel of each stream associated in the subset of block, the block in the subset considered is decoded according to the predefined procedure of traversal, and for the current block of considered subset, described decoding device comprises:
Based at least one symbol appearing probability, current block is carried out to the subassembly of entropy decoding,
About the block of decoding before at least one, current block is carried out to the subassembly of prediction decoding,
The described decoding device part that merits attention is:
When-will decode in the subset that current block is considered first piece, the entropy for the first current block is decoded, and the subassembly carrying out entropy decoding considers the symbol appearing probability calculated for the predetermined block of decoding of at least one other subset,
-when being block in the subset of described consideration except first piece at current block, the entropy for current block is decoded, and the subassembly carrying out entropy decoding considers the symbol appearing probability calculated at least one decoding block belonging to same subsets.
Target of the present invention is also a kind of computer program comprising instruction, and when this program performs on computers, described instruction is for performing above coding or the step of coding/decoding method.
This program can use any programming language, and can be the form of source code, object code or the intermediate code between source code and object code, the form of such as partial compilation, or other forms wanted arbitrarily.
Another target of the present invention is also a kind of recording medium that can be read by computer, and it comprises such as above computer program command.
Recording medium can be can stored program any entity or equipment.Such as, this medium can comprise: storage device, such as ROM (as CD ROM) or microelectronic circuit ROM; Or magnetic recording system, such as disk (floppy disk) or hard disk.
In addition, this recording medium can be transmission medium such as electricity or light signal, and it can be transmitted with radio or other modes by electricity or optical cable.Can download on the network of Internet type especially according to program of the present invention.
Alternatively, this recording medium can be integrated circuit, and program is incorporated in this integrated circuit, and this circuit is suitable for performing the method discussed or the execution that can be used for the latter.
Above-mentioned encoding device, coding/decoding method, decoding device and computer program at least have the Dominant Facies advantage together showed with coding method according to the present invention.
Accompanying drawing explanation
By reference to two preferred embodiments that accompanying drawing describes, other Characteristics and advantages will become obvious, in the accompanying drawings:
-Figure 1A represents the Image Coding figure of the prior art according to the first example,
-Figure 1B represents the Image Coding figure of the prior art according to second example,
-Fig. 2 A represents the key step according to coding method of the present invention,
-Fig. 2 B illustrates the parallel encoding realized in coding method in fig. 2 in detail,
-Fig. 3 A represents the embodiment according to encoding device of the present invention,
-Fig. 3 B represents the coding unit of the encoding device in Fig. 3 A,
-Fig. 4 A represents the encoding/decoding image figure according to the first preferred embodiment,
-Fig. 4 B represents the encoding/decoding image figure according to the second preferred embodiment,
-Fig. 5 A represents the key step according to coding/decoding method of the present invention,
-Fig. 5 B illustrates the parallel encoding realized in the coding/decoding method in Fig. 5 A in detail,
-Fig. 6 A represents the embodiment according to decoding device of the present invention,
-Fig. 6 B represents the decoding unit of the decoding device in Fig. 6 A.
Embodiment
Embodiment of the present invention will be described now, and wherein, coding method according to the present invention is used to encode to a series of images according to binary stream, and this binary stream is close to the binary stream obtained according to the coding of H.264/MPEG-4AVC standard.In this embodiment, by such as realizing according to coding method of the present invention in the mode of software or hardware the adjustment of the encoder initially meeting H.264/MPEG-4AVC standard.Show according to coding method of the present invention with the form comprising the algorithm of step C1 to C5 shown in Fig. 2 A.
According to embodiments of the invention, realize according to coding method of the present invention in the encoding device CO shown in Fig. 3 A.
With reference to figure 2A, the first coding step C1 is the image IE cutting in a series of images that will encode is multiple pieces or macro block MB, as shown in Fig. 4 A or 4B.In the example shown, described piece of MB is square and has identical size.As the function of image size, it needs not to be the multiple of block size, and the last block in left side and the last block of bottom can not be squares.In alternate embodiments, block can be such as rectangle size and/or not line up mutually.
Each piece or macro block self also can be divided into sub-block, and this sub-block also can be segmented self.
This cutting is realized by the division module PCO shown in Fig. 3 A, its example partitioning algorithm as everyone knows.
With reference to figure 2A, the second coding step C2 be the continuous print block above-mentioned piece being grouped into predetermined quantity P subset SE1, SE2 ... SEk ..., SEP, this subset will by parallel encoding.In the example shown in Fig. 4 A and 4B, predetermined quantity P equals 4, and represented by dashed line four subset SE1, SE2, SE3, SE4 are made up of four lines block before image IE respectively.
This division operation utilizes well-known algorithm to realize by the computing module GRCO shown in Fig. 3 A.
With reference to figure 2A, the 3rd coding step C3 comprises each parallel encoding in subset SE1 to described piece, SE2, SE3 and SE4, and the block in the subset considered is encoded according to the predefined procedure PS of traversal.In the example shown in Fig. 4 A and 4B, the block in current subnet SEk (1≤k≤4) is from left to right encoded successively as shown in arrow P S.
This parallel encoding is that the coding unit UCk (1≤k≤R) being R (R=4) by the quantity shown in Fig. 3 A realizes, and allows the significantly acceleration of coding method.By mode known like this, encoder CO comprises buffer storage MT, its be suitable for comprising such as with the symbol appearing probability upgraded gradually of the coding cooperative of current block.
As Fig. 3 B shows in more detail, each coding unit UCk comprises:
About the block of Code And Decode before at least one, current block is carried out to the subelement of predictive coding, represent with SUCPk;
At least one symbol appearing probability calculated by the block used for described Code And Decode before carries out the subelement of entropy code to described current block, represent with SUCEk.
Predictive coding subelement SUCPk can according to traditional Predicting Technique such as with in frame and/or inter-frame mode perform the predictive coding of current block.
Entropy code subelement SUCEk itself is CABAC type, but adjusts according to the present invention, will further describe in the description.
As variant, entropy code subelement SUCEk can be known huffman encoder.
In the example shown in Fig. 4 A and 4B, the block of first module UC1 to the first row SE1 is encoded from left to right.When arriving last block of the first row SE1, first piece that (N+1) row is the 5th row etc. here can be delivered to.The block of second unit UC2 to the second row SE2 is encoded from left to right.When last block of arrival second row SE2, first piece that (N+2) row is the 6th row etc. here can be delivered to.This traversal is repeated until unit UC4, and it is encoded from left to right to the block of fourth line SE4.When arriving last block of the first row, first piece that (N+4) row is eighth row here can be delivered to, by that analogy until last block of image IE is encoded.
Certainly be also possible from the traversal of the different other types just now described above.So, can be several subgraphs by image IE cutting, and the cutting of the type is applied to each subgraph independently.Each coding unit also can not process row nested as explained above but nested row.Row or column can also be traveled through with any direction.
With reference to figure 2A, the 4th coding step C4 produces the decoded version of processed piece in N number of sub-bit stream Fk (1≤k≤N) and each subset SEk, and this sub-bit stream Fk represents by the processed block of above-mentioned each coding unit compression.According to the synchronization mechanism that will describe in further detail in the description, with SED1, SED2 ... SEDk ..., processed piece of decoding in the subset considered that represents of SEDP can by some coding unit UC1 shown in Fig. 3 A, UC2 ..., UCk ..., UCP reuses.
With reference to figure 2A, the 5th coding step C5 comprises the stream F constructing the overall situation based on above-mentioned subflow Fk.According to an embodiment, subflow Fk is by arranged side by side simply, and its extra item of information is used for representing the position of each subflow Fk in global flow F to decoder.Described global flow F is transferred to remote terminal by communication network (not shown).The latter comprises the decoder DO shown in Fig. 6 A.
So, as described in more detail in specification, the subflow Fk in global flow F can be isolated according to decoder of the present invention, and they be distributed to each composition decoding unit of decoder.It is noted that and subflow is decomposed into the global flow decoding unit that several operate concurrently independent of choice for use, and only can be used the encoder or decoder that comprise parallel operated units by the method.
This structure of global flow F realizes in stream constructing module CF as shown in Figure 3A.
Each particular child step of the present invention such as realized in coding unit UCk during above-mentioned parallel encoding step C3 is described referring now to Fig. 2 B.
In the process of step C31, coding unit UCk select will to encode in the current line SEk shown in Fig. 4 A or 4B first piece is as current block.
In the process of step C32, whether unit UCk tests current block is first piece (being positioned at top and left side) in image IE, and in above-mentioned steps C1, this image has been split as block.
If this situation, in the process of step C33, the value Pinit defined in the encoder CO of Fig. 3 A before coding possibilities is initialized to.
If not this situation, by the step C40 of detailed description in follow-up explanation, determine the availability of the necessary block of Code And Decode before.
In the process of step C34, the first current block MB1 in the first row SE1 described in Fig. 4 A or 4B is encoded.This step C34 comprises multiple sub-step C341 to the C348 that will be described below.
In the process of the first sub-step C341, by frame and/or the known technology of inter prediction predictive coding is carried out to current block MB1, in its process, predict block MB1 about the block of Code And Decode before at least one.
Needless to say, the infra-frame prediction of other patterns such as proposed in the h .264 standard is also fine.
Current block MB1 through the predictive coding of inter-frame mode, in its process, can also predict current block about the block from the image generation of Code And Decode before.Certainly it is contemplated that the prediction of other types.For in the possible prediction of current block, select optimum prediction according to well-known rate distortion criterion to those skilled in the art.
Above-mentioned predictive coding step makes may structure forecast block MBp
1, this is to current block MB
1approximate.The information relevant to this predictive coding will be written to subsequently and flow F, and this stream F is sent to decoder DO.This information comprises the type (interframe or frame in) of prediction especially, if and suitable, the offset vector comprising the classified types (if macro block is subdivided) of the pattern of infra-frame prediction, block or macro block, reference picture index and use in inter-frame forecast mode.This information is compressed by encoder CO.
In the process of sub-step C342 below, from current block MB
1in deduct prediction block MBp
1, produce residual block MBr
1.
In the process of sub-step C343 below, convert residual block MBr according to traditional Direct Transform operation (such as the discrete cosine transform of DCT type)
1, produce the block MBt of conversion
1.
In the process of sub-step C344 below, carry out the block MBt to conversion according to traditional quantization operation (such as scalar quantization)
1quantize.Then the block MBq that coefficient is quantized is obtained
1.
In the process of sub-step C345 below, to the block MBq that coefficient is quantized
1carry out entropy code.In a preferred embodiment, this needs CABAC entropy code.
In the process of sub-step C346 below, come block MBq according to traditional inverse quantization operation
1carry out inverse quantization, this is the inverse operation of the quantification performed in step C344.Then coefficient is obtained by the block MBDq of inverse quantization
1.
In the process of sub-step C347 below, to coefficient by the block MBDq of inverse quantization
1carry out inverse transformation, this is the inverse operation of the Direct Transform performed in above step C343.Then the residual block MBDr of decoding is obtained
1.
In the process of sub-step C348 below, by the residual block MBDr by decoding
1be added to prediction block MBp
1construct the block MBD of decoding
1.It is noted that the decoding block that the latter obtains when completing with the method for decoding to image IE further described in the description is identical.Decoding block MBD
1so can be used by other coding units any of coding unit UC1 or the coding unit forming predetermined quantity R.
When completing above-mentioned coding step C34, such as entropy code subelement SUCEk as shown in Figure 3 B comprises such as with the coding cooperative of first piece and all probability upgraded gradually.These probability correspond to various possible syntactic element and correspond to various relevant coding context.
After above-mentioned coding step C34, test in the process of step C35, to determine whether current block is the jth block that this phase is gone together, and wherein, j is the known predetermined value of encoder CO and at least equals 1.
If such situation, in the process of step C36, the one group of probability calculated for jth block is stored in the buffer storage MT of the encoder CO such as shown in Fig. 3 A and Fig. 4 A and 4B, and the size of described memory is suitable for the probability storing the quantity calculated.
In the process of step C37, unit UCk tests just now by last block that whether current block in the row SEk that encodes is image IE.
If this situation, then, in the process of step C38, coding method terminates.
If not this situation, the block MB that the traversal order in the process of step C39 represented by the arrow P S in Fig. 4 A or 4B selects next to encode
i.
If in the process of step C35, current block is not the jth block in considered row SEk, then without above step C37.
In the process of step C40, determine for current block MB
icoding necessary before the availability of block of Code And Decode.Consider that this needs to be come by different coding unit UCk the fact of the block parallel encoding to image IE, may not carry out Code And Decode to these blocks for the coding unit of encoding to these blocks, and therefore they are also unavailable.Described determining step comprises block that checking is arranged in the predetermined quantity N' of previous row SEk-1 and such as lays respectively at the coding whether top of current block and top-right two blocks can be used for current block, and namely whether they have been encoded by the coding unit UCk-1 encoded to it and then decoded.Described determining step also comprises checking and is positioned at the current block MB that will encode
ithe availability of at least one piece in left side.But, consider the traversal order PS selected in the embodiment shown in Fig. 4 A or 4B, the block in considered row SEk encoded successively.Therefore, the block of left side encoded and decoding always can (except first in a line piece).At the example shown in Fig. 4 A or 4B, this needs block to be disposed immediately in the left side of the current block that will encode.For this reason, only test lays respectively at the top of current block and the availability of top-right two blocks.
This testing procedure is easy to make coding method slack-off, by alternative method according to the present invention, clock CLK shown in Fig. 3 A is suitable for carrying out synchronously to the process of block forecast, thus guarantee to lay respectively at the top of current block and the availability of top-right two blocks, and do not need the availability verifying these two blocks.So as shown in Fig. 4 A or 4B, coding unit UCk is always for the predetermined quantity N'(N'=2 here in the previous row SEk-1 of the coding of current block) Code And Decode block as offseting the coding that starts first piece.From the angle of software, realize such clock and make significantly to accelerate for the treatment of the processing time needed for the block in image IE in encoder CO.
In the process of step C41, carry out testing determining that whether current block is first piece in considered row SEk.
If such situation, in the process of step C42, the symbol appearing probability calculated only read in the coding of the jth block of previous row SEk-1 in buffer storage MT during.
The first variant according to Fig. 4 A, jth block is first piece (j=1) in previous row SEk-1.This reading comprises the probability replacing CABAC encoder with the probability existed in buffer storage MT.Process each in the second row SE2, the third line SE32 and fourth line SE4 first piece, the arrow represented by fine rule in Figure 4 A describes this read step.
Second variant of the above-mentioned steps C42 according to Fig. 4 B, jth block is second piece (j=2) of previous row SEk-1.This reading comprises the probability replacing CABAC encoder with the probability existed in buffer storage MT.Process each in the second row SE2, the third line SE32 and fourth line SE4 first piece, the arrow represented by a dotted line is in figure 4b to describe this read step.
After step C42, by the iteration of above-mentioned steps C34 to C38, current block is encoded and then decoded.
If after above-mentioned steps C41, current block is not first piece of considered row SEk, advantageously can not read from be arranged in same a line SEk before the probability that produces of the block (namely shown example is disposed immediately in the Code And Decode block in the left side of current block) of Code And Decode.Really, consider that the order traversal of the block being arranged in same a line reads PS, as shown in Fig. 4 A or 4B, the symbol appearing probability that the symbol appearing probability existed in CABAC encoder when the coding of current block starts exists after being just in time the coding/decoding of last piece in this phase colleague.
Therefore, in the process of step C43, learn the symbol appearing probability of the entropy code of described current block, its only correspond to as the double solid line arrow in Fig. 4 A or 4B represent for the last piece of symbol appearing probability calculated as described in same a line.
After step C43, by the iteration of above-mentioned steps C34 to C38, current block is encoded and then decoded.
The detailed description of the embodiment of decoded portion
Embodiment according to coding/decoding method of the present invention will be described now, wherein, in the mode of software or hardware by realizing this coding/decoding method to the adjustment of the decoder initially meeting H.264/MPEG-4AVC standard.
Show according to coding/decoding method of the present invention with the form comprising the algorithm of step D1 to D4 shown in Fig. 5 A.
According to embodiments of the invention, realize according to coding/decoding method of the present invention in decoding device DO as shown in Figure 6A.
With reference to figure 5A, the first decoding step D1 be identify in described stream F N number of subflow F1, F2 ..., Fk ..., FP, this subflow comprise respectively as shown in Fig. 4 A or 4B before coding block or macro block MB N number of subset SE1, SE2 ..., SEk ..., SEP.For this reason, each subflow Fk in stream F associates with designator, determines the position of each subflow in stream F to allow decoder DO.In the example shown, described piece of MB has square and has identical size.As the function of image size, it needs not to be the multiple of block size, and the last block in left side and the last block of bottom can not be squares.In alternate embodiments, block can be such as rectangle size and/or not line up mutually.
Each piece or macro block self also can be divided into sub-block, and this sub-block also can be segmented self.
Such identification is that the stream extraction module EXDO by such as representing in fig. 6 performs.
In the example shown in Fig. 4 A or 4B, predetermined quantity equals 4, and four subset SE1, SE2, SE3, SE4 are represented by dashed line.
With reference to figure 5A, the second decoding step D2 is each parallel decoding in subset SE1 to described piece, SE2, SE3 and SE4, and the block in the subset considered is encoded according to the order PS of traversal.In the example shown in Fig. 4 A or 4B, the block in current subnet SEk (1≤k≤4) is from left to right decoded successively as shown in arrow P S.When step D2 completes, obtain decoding block subset SED1, SED2, SED3 ..., SEDk ..., SEDP.
This parallel decoding is that the decoding unit UDk (1≤k≤R) being R (R=4) by quantity as shown in Figure 6A realizes, and allows the significantly acceleration of coding/decoding method.According to mode known like this, decoder DO comprises buffer storage MT, and it is suitable for comprising the symbol appearing probability such as upgraded gradually collaboratively with the decoding of current block.
As Fig. 6 B shows in more detail, each decoding unit UDk comprises:
By learning at least one symbol appearing probability calculated for the block of decoding before at least one, the subelement described current block being carried out to entropy decoding, represent with SUDEk,
About described block of decoding before, current block is carried out to the subelement of prediction decoding, represent with SUDPk.
Prediction decoding subelement SUDPk such as can perform the prediction decoding of current block in frame and/or under inter-frame mode according to traditional Predicting Technique.
Entropy decoding subelement SUDEk itself is CABAC type, but adjusts according to the present invention, will further describe in the description.
As variant, entropy decoding subelement SUDEk can be known huffman decoder.
In the example shown in Fig. 4 A or 4B, first module UD1 decodes from left to right to the block in the first row SE1.When arriving last block of the first row SE1, (N+1) row (being the 5th row) etc. first piece can be delivered to here.Second unit UC2 from left to right decodes to the block in the second row SE2.When last block of arrival second row SE2, (N+2) row (being the 6th row) etc. first piece can be passed to here.This traversal repeats until unit UD4, and it is decoded from left to right to the block in fourth line SE4.When arriving last block of the first row, can pass to (N+4) row (being eighth row) first piece here, last block in the subflow to the last identified by that analogy is decoded.
Certainly be also possible from the traversal of the different other types just now described above.Such as, each decoding unit also can not process row nested as explained above but nested row.Row or column can also be traveled through with any direction.
With reference to figure 5A, the 3rd decoding step D3 be based on each decoding obtained in decoding step D2 subset SED1, SED2 ..., SEDk ..., SEDP reconstructs the image of decoding.More precisely, each decoding subset SED1, SED2 ..., SEDk ..., decoding block in SEDP is sent to such as image reconstruction unit URI as shown in Figure 6A.In the process of step D3, when these blocks become available, unit URI writes decoding block in the image of decoding.
In the process of the 4th decoding step D4 shown in Fig. 5 A, the unit URI shown in Fig. 6 A provides the image ID of complete decoding.
Each particular child step of the present invention such as realized in decoding unit UDk during above-mentioned parallel decoding step D2 is described referring now to Fig. 5 B.
In the process of step D21, decoding unit UDk select will to decode in the current block SEk shown in Fig. 4 A or 4B first piece is as current block.
In the process of step D22, whether unit UDk tests current block is in decoded image first piece, is first piece of subflow F1 in this example.
If this situation, in the process of step D23, the value Pinit defined in the decoder DO of Fig. 6 A before decoding probability is initialized to.
If not this situation, by the availability of the block of decoding before determining necessity in the process of the step D30 described in follow-up explanation.
In the process of step D24, the first current block MB1 in the first row SE1 shown in Fig. 4 A or 4B is decoded.This step D24 comprises following multiple sub-step D241 to the D246 that will describe.
In the process of the first sub-step D241, entropy decoding is carried out to the syntactic element relevant to current block.More precisely, decoded by the syntactic element that CABAC entropy decoding SUDE1 pair, the subelement such as shown in Fig. 6 B is relevant to current block.The sub-bit stream F1 of CABAC entropy decoding subelement SUDE1 to compressed file decodes to produce syntactic element, and upgrade its probability in like fashion simultaneously, thus when this subelement is decoded to symbol, the probability of occurrence of this symbol equals the symbol appearing probability obtained when encoding to this same-sign in above-mentioned entropy code step C345.
In the process of sub-step D242 below, by frame and/or the known technology of inter prediction prediction decoding is carried out to current block MB1, in its process, the block about decoding before at least one predicts described piece of MB1.
Needless to say, the infra-frame prediction of other patterns such as proposed in the h .264 standard is also fine.
In the process of this step, prediction decoding is carried out by means of the syntactic element of decoding in step before, this syntactic element comprises the type (interframe or frame in) of prediction especially, if and suitable, the offset vector comprising the classified types (if the latter is subdivided) of the pattern of infra-frame prediction, block or macro block, reference picture index and use in inter-frame forecast mode.
Above-mentioned prediction decoding step makes may structure forecast block MBp
1.
In the process of sub-step D243 below, construct by means of the syntactic element of decoding before and quantize residual block MBq
1.
In the process of sub-step D244 below, carry out the residual block MBq to quantizing according to traditional inverse quantization operation
1carry out inverse quantization, this is the inverse operation of the quantification performed in above-mentioned steps C344, to produce the inverse quantisation block MBDt of decoding
1.
In the process of sub-step D245 below, to inverse quantisation block MBDt
1carry out inverse transformation, this is the inverse operation of the Direct Transform performed in above step C343.Then the residual block MBDr of decoding is obtained
1.
In the process of sub-step D246 below, by block MBp will be predicted
1be added to the residual block MBDr of decoding
1carry out construction solution code block MBD
1.Described decoding block MBD
1so can other decoding units any of a part of decoded unit UD1 or the decoding unit that forms predetermined quantity N use.
When completing above-mentioned decoding step D246, such as entropy decoding subelement SUDE1 as shown in Figure 6B comprises and such as cooperates with the decoding of first piece and all probability of upgrading gradually.These probability correspond to the decoding context of various possible syntactic element and various association.
After above-mentioned decoding step D24, test in the process of step D25, to determine whether current block is the jth block that this phase is gone together, and wherein j is the known predetermined value of decoder DO and at least equals 1.
If this situation, in the process of step D26, the one group of probability calculated for jth block is stored in the buffer storage MT of the decoder DO such as shown in Fig. 6 A and Fig. 4 A or 4B, and the size of shown memory is suitable for the quantity storing the probability calculated.
In the process of step D27, whether unit UDk tests just now decoded current block is last block in last subflow.
If this situation, in the process of step D28, coding/decoding method terminates.
If not this situation, perform step D29, the traversal order in the process of step D29 according to the arrow P S in Fig. 4 A or 4B selects next block MB that will decode
i.
If in the process of above-mentioned steps D25, current block is not the jth block in considered row SEDk, carries out above step D27.
In the process of the step D30 after above-mentioned steps D29, determine for current block MB
idecoding necessary before the availability of block of decoding.Consider that these needs carry out the fact to block parallel encoding by different decoding unit UDk, may not decode to these blocks for the decoding unit of decoding to these blocks, and therefore they are also unavailable.Described determining step comprises verifies whether the block (such as laying respectively at the top of current block and top-right two blocks) being arranged in the predetermined quantity N' of previous row SEk-1 can be used for the decoding of current block, and namely whether they are decoded by the decoding unit UDk-1 decoded to it.Described determining step also comprises checking and is positioned at the current block MB that will decode
ithe availability of at least one piece in left side.But, consider the traversal order PS selected in the embodiment shown in Fig. 4 A or 4B, the block in considered row SEk decoded successively.Therefore, the block of the decoding in left side always can (except first in a line piece).In the example shown in Fig. 4 A or 4B, this needs block to be disposed immediately in the left side of the current block that will decode.For this reason, only test lays respectively at the top of current block and the availability of top-right two blocks.
This testing procedure is easy to make coding/decoding method slack-off, by alternative method according to the present invention, clock CLK shown in Fig. 6 A is suitable for carrying out synchronously to the process of block decoding, thus guarantee to lay respectively at the top of current block and the availability of top-right two blocks, and do not need the availability verifying these two blocks.So, as shown in Fig. 4 A or 4B, decoding unit UDk always for decoding current block previous row SEk-1 in predetermined quantity N'(N'=2 here) decoding block as skew start to decode to first piece.From the angle of software, realize such clock and make significantly to accelerate to process the processing time needed for block in each subset SEk in decoder DO.
In the process of step D31, carry out testing determining that whether current block is first piece in considered row SEk.
If such situation, in the process of step D32, the symbol appearing probability calculated only read in the decoding of the jth block of previous row SEk-1 in buffer storage MT during.
The first variant according to Fig. 4 A, jth block is first piece (j=1) in previous row SEk-1.This reading comprises the probability replacing CABAC decoder with the probability existed in buffer storage MT.Process equally each in the second row SE2, the third line SE3 and fourth line SE4 first piece, the arrow represented by fine rule in Figure 4 A describes this read step.
Second variant of the above-mentioned steps D32 according to Fig. 4 B, jth block is second piece (j=2) of previous row SEk-1.This reading comprises the probability replacing CABAC decoder with the probability existed in buffer storage MT.Process each in the second row SE2, the third line SE3 and fourth line SE4 first piece, the arrow represented by a dotted line is in figure 4b to describe this read step.
After step D32, by the iteration of above-mentioned steps D24 to D28, current block is decoded.
If after above-mentioned steps D31, current block is not first piece of considered row SEk, advantageously can not read from be arranged in same a line SEk before the probability that produces of the block (namely shown example is disposed immediately in the decoding block in the left side of current block) of decoding.Really, consider that the order traversal of the block being arranged in same a line reads PS, as shown in Fig. 4 A or 4B, the symbol appearing probability that the symbol appearing probability existed in CABAC decoder when the decoding of current block starts exists after being just in time the decoding of last piece in this phase colleague.
Therefore, in the process of step D33, learn the symbol appearing probability that the entropy for described current block is decoded, its only correspond to as the double solid line arrow in Fig. 4 A or 4B represent for the last piece of symbol appearing probability calculated as described in same a line.
After step D33, by the iteration of above-mentioned steps D24 to D28, current block is decoded.
Claims (14)
1. storage comprises a computer-readable medium for the software of the instruction that can be performed by one or more computer, and described instruction makes one or more computer described perform the operation comprising following item when being performed:
Receive the stream representing at least one coded image;
The group of predetermined multiple pieces of representative at least one coded image described is identified from stream;
In the group of described piece each group is supplied to the different decoding units in multiple decoding units of parallel work-flow; And
By first piece in the group of the given block of the first decoding unit process, wherein the process of first piece is comprised:
From the buffer shared by described multiple decoding unit, fetch first group of probability data, wherein first group of probability data comprise first group with the described multiple pieces symbol appearing probabilities be associated;
Entropy decoding is carried out to first piece based on first group of probability data;
By second piece in the group of block given described in the first decoding unit process, wherein the process of second piece is comprised:
From the memory cell be associated with the first decoding unit, fetch second group of probability data, wherein second group of probability data comprise second group with the described multiple pieces symbol appearing probabilities be associated; And
Entropy decoding is carried out to second piece based on second group of probability data.
2. computer-readable medium as claimed in claim 1, wherein, described operation comprises further:
The 3rd group of probability data is generated based on second group of probability data with about the data of second piece; And
Second group of probability data in the memory cell be associated with the first decoding unit is replaced with the 3rd group of probability data.
3. computer-readable medium as claimed in claim 1, wherein:
Carry out entropy decoding based on first group of probability data to first piece to comprise: by using context to adapt to binary arithmetic coding (CABAC) algorithm, carry out entropy decoding based on first group of probability data to first piece; And
Carry out entropy decoding based on second group of probability data to second piece to comprise: by using CABAC algorithm, carry out entropy decoding based on second group of probability data to second piece.
4. computer-readable medium as claimed in claim 1, wherein, described operation comprises further:
The 3rd group of probability data is generated based on first group of probability data with about the data of first piece;
3rd group of probability data is stored into the memory cell be associated with the first decoding unit; And
At least the 3rd group of probability data is used to generate second group of probability data.
5. computer-readable medium as claimed in claim 1, wherein, described operation comprises further:
Relative to the block of at least one early decoding, prediction decoding is carried out to each piece in described multiple pieces.
6. computer-readable medium as claimed in claim 1, wherein, described operation comprises further:
By the 3rd piece in the group of block given described in the first decoding unit process, wherein the process of the 3rd piece is comprised:
The 3rd group of probability data is fetched from the memory cell be associated with the first decoding unit, wherein the 3rd group of probability data comprise the 3rd group with the described multiple pieces symbol appearing probabilities be associated, and wherein the 3rd group of probability data at least based on second group of probability data; And
Entropy decoding is carried out to the 3rd piece based on the 3rd group of probability data.
7. computer-readable medium as claimed in claim 1, wherein, described operation comprises further:
Using the different decoding units in the described multiple decoding unit of predetermined time interval selection as selected decoding unit;
One group of new probability data is provided to the buffer shared by described multiple decoding unit by the decoding unit selected; And
The all data be stored in the buffer shared by described multiple decoding unit are replaced with new one group of probability data.
8. a system, comprising:
The memory device of one or more computer and one or more storage instruction, described instruction being operable is made one or more computer described perform the operation comprising following item when being performed by one or more computer described:
Receive the stream representing at least one coded image;
The group of predetermined multiple pieces of representative at least one coded image described is identified from stream;
In the group of described piece each group is supplied to the different decoding units in multiple decoding units of parallel work-flow; And
By first piece in the group of the given block of the first decoding unit process, wherein the process of first piece is comprised:
From the buffer shared by described multiple decoding unit, fetch first group of probability data, wherein first group of probability data comprise first group with the described multiple pieces symbol appearing probabilities be associated;
Entropy decoding is carried out to first piece based on first group of probability data;
By second piece in the group of block given described in the first decoding unit process, wherein the process of second piece is comprised:
From the memory cell be associated with the first decoding unit, fetch second group of probability data, wherein second group of probability data comprise second group with the described multiple pieces symbol appearing probabilities be associated; And
Entropy decoding is carried out to second piece based on second group of probability data.
9. system as claimed in claim 8, wherein, described operation comprises further:
The 3rd group of probability data is generated based on second group of probability data with about the data of second piece; And
Second group of probability data in the memory cell be associated with the first decoding unit is replaced with the 3rd group of probability data.
10. system as claimed in claim 8, wherein:
Carry out entropy decoding based on first group of probability data to first piece to comprise: by using context to adapt to binary arithmetic coding (CABAC) algorithm, carry out entropy decoding based on first group of probability data to first piece; And
Carry out entropy decoding based on second group of probability data to second piece to comprise: by using CABAC algorithm, carry out entropy decoding based on second group of probability data to second piece.
11. systems as claimed in claim 8, wherein, described operation comprises further:
The 3rd group of probability data is generated based on first group of probability data with about the data of first piece;
3rd group of probability data is stored into the memory cell be associated with the first decoding unit; And
At least the 3rd group of probability data is used to generate second group of probability data.
12. systems as claimed in claim 8, wherein, described operation comprises further:
Relative to the block of at least one early decoding, prediction decoding is carried out to each piece in described multiple pieces.
13. systems as claimed in claim 8, wherein, described operation comprises further:
By the 3rd piece in the group of block given described in the first decoding unit process, wherein the process of the 3rd piece is comprised:
The 3rd group of probability data is fetched from the memory cell be associated with the first decoding unit, wherein the 3rd group of probability data comprise the 3rd group with the described multiple pieces symbol appearing probabilities be associated, and wherein the 3rd group of probability data at least based on second group of probability data; And
Entropy decoding is carried out to the 3rd piece based on the 3rd group of probability data.
14. systems as claimed in claim 8, wherein, described operation comprises further:
Using the different decoding units in the described multiple decoding unit of predetermined time interval selection as selected decoding unit;
One group of new probability data is provided to the buffer shared by described multiple decoding unit by the decoding unit selected; And
The all data be stored in the buffer shared by described multiple decoding unit are replaced with new one group of probability data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510340428.5A CN105007486B (en) | 2011-03-07 | 2012-02-23 | Code and decode method, the coding and decoding equipment of image |
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR1151849A FR2972588A1 (en) | 2011-03-07 | 2011-03-07 | METHOD FOR ENCODING AND DECODING IMAGES, CORRESPONDING ENCODING AND DECODING DEVICE AND COMPUTER PROGRAMS |
| FR11/51,849 | 2011-03-07 | ||
| CN201510340428.5A CN105007486B (en) | 2011-03-07 | 2012-02-23 | Code and decode method, the coding and decoding equipment of image |
| CN201280021355.8A CN103535032B (en) | 2011-03-07 | 2012-02-23 | Coding and the decoding method of image, coding and decoding device |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201280021355.8A Division CN103535032B (en) | 2011-03-07 | 2012-02-23 | Coding and the decoding method of image, coding and decoding device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105007486A true CN105007486A (en) | 2015-10-28 |
| CN105007486B CN105007486B (en) | 2018-06-05 |
Family
ID=54447112
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510340428.5A Active CN105007486B (en) | 2011-03-07 | 2012-02-23 | Code and decode method, the coding and decoding equipment of image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105007486B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110291793A (en) * | 2017-01-06 | 2019-09-27 | 联发科技股份有限公司 | The method and apparatus that range derives in context adaptive binary arithmetic coding/decoding |
| US11425386B2 (en) | 2017-07-14 | 2022-08-23 | Hfi Innovation Inc. | Method and apparatus for range derivation in context adaptive binary arithmetic coding |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008036237A2 (en) * | 2006-09-21 | 2008-03-27 | Analog Devices, Inc. | Multiprocessor decoder system and method |
| US20090168868A1 (en) * | 2007-12-31 | 2009-07-02 | Musa Jahanghir | Systems and apparatuses for performing CABAC parallel encoding and decoding |
| CN101490968A (en) * | 2006-07-17 | 2009-07-22 | 索尼株式会社 | Parallel processing apparatus for video compression |
| CN101836454A (en) * | 2008-12-03 | 2010-09-15 | 联发科技股份有限公司 | Method for performing parallel cabac processing with ordered entropy slices, and associated apparatus |
-
2012
- 2012-02-23 CN CN201510340428.5A patent/CN105007486B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101490968A (en) * | 2006-07-17 | 2009-07-22 | 索尼株式会社 | Parallel processing apparatus for video compression |
| WO2008036237A2 (en) * | 2006-09-21 | 2008-03-27 | Analog Devices, Inc. | Multiprocessor decoder system and method |
| US20090168868A1 (en) * | 2007-12-31 | 2009-07-02 | Musa Jahanghir | Systems and apparatuses for performing CABAC parallel encoding and decoding |
| CN101836454A (en) * | 2008-12-03 | 2010-09-15 | 联发科技股份有限公司 | Method for performing parallel cabac processing with ordered entropy slices, and associated apparatus |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110291793A (en) * | 2017-01-06 | 2019-09-27 | 联发科技股份有限公司 | The method and apparatus that range derives in context adaptive binary arithmetic coding/decoding |
| CN110291793B (en) * | 2017-01-06 | 2022-01-25 | 联发科技股份有限公司 | Method and apparatus for range derivation in context adaptive binary arithmetic coding |
| US11265561B2 (en) | 2017-01-06 | 2022-03-01 | Mediatek Inc. | Method and apparatus for range derivation in context adaptive binary arithmetic coding |
| US11425386B2 (en) | 2017-07-14 | 2022-08-23 | Hfi Innovation Inc. | Method and apparatus for range derivation in context adaptive binary arithmetic coding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105007486B (en) | 2018-06-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| RU2766445C2 (en) | Method for image encoding and decoding, encoding and decoding device and corresponding computer programs | |
| CN105120277A (en) | Method for encoding and decoding images, encoding and decoding device, and corresponding computer programs | |
| CN102783035A (en) | Parallel entropy coding and decoding methods and devices | |
| CN105007486A (en) | Method of coding and decoding images, coding and decoding device and computer programs |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1214050 Country of ref document: HK |
|
| CB02 | Change of applicant information |
Address after: Amsterdam Applicant after: Dolby International AB Address before: Stockholm Applicant before: Dolby International AB |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |