WO2007081116A1 - Procédé et appareil de codage et de décodage de données vidéo - Google Patents

Procédé et appareil de codage et de décodage de données vidéo Download PDF

Info

Publication number
WO2007081116A1
WO2007081116A1 PCT/KR2007/000082 KR2007000082W WO2007081116A1 WO 2007081116 A1 WO2007081116 A1 WO 2007081116A1 KR 2007000082 W KR2007000082 W KR 2007000082W WO 2007081116 A1 WO2007081116 A1 WO 2007081116A1
Authority
WO
WIPO (PCT)
Prior art keywords
residual
vector
vectors
transform coefficient
path
Prior art date
Application number
PCT/KR2007/000082
Other languages
English (en)
Inventor
Se-Yoon Jeong
Jeong-Il Seo
Kyu-Heon Kim
Kyeongok Kang
Jin-Woo Hong
Yung-Lyul Lee
Dae-Yeon Kim
Dong-Gyun Kim
Seoung-Jun Oh
Dong-Gyu Sim
Chang-Beom Ahn
Original Assignee
Electronics And Telecommunications Research Institute
Kwangwoon University Research Institute For Industry Cooperation
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Electronics And Telecommunications Research Institute, Kwangwoon University Research Institute For Industry Cooperation filed Critical Electronics And Telecommunications Research Institute
Priority to US12/160,154 priority Critical patent/US20090067503A1/en
Priority claimed from KR1020070001386A external-priority patent/KR100867995B1/ko
Publication of WO2007081116A1 publication Critical patent/WO2007081116A1/fr

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/94—Vector quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H04N19/107—Selection of coding mode or of prediction mode between spatial and temporal predictive coding, e.g. picture refresh
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/18—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Definitions

  • the present invention relates to video data encoding and decoding methods and apparatuses, and more particularly, to video data encoding and decoding methods and apparatuses capable of increasing a compression ratio and reducing calculation complexity in a decoder by encoding only an index corresponding to a vector that is most similar to a residual transform coefficient among vectors of a code book in an operation of encoding the residual transform coefficient obtained by performing discrete cosine transformation (DCT) and quantization.
  • DCT discrete cosine transformation
  • AVC advanced video coding
  • CABAC Context Adaptive Binary Arithmetic Coding
  • the highest probability appears in the left upper region (DC), and a distribution of probabilities has symmetry with respect to horizontal and vertical directions.
  • the zigzag scan is carried out from the DC residual transform coefficient, so that the residual transform coefficients are reordered into a 1-D array. Since the reordered residual transform coefficients include a large number of zeros, "0", the residual transform coefficients can be represented in a simpler form by using the Run-Level encoding procedure.
  • Run and Level denotes the size of a non-zero residual transform coefficient.
  • the present invention provides video data encoding method and an apparatus for encoding a residual transform coefficient obtained by performing discrete cosine transformation (DCT) or other types of transformation and quantization, capable of increasing compression ratio by encoding only an index corresponding to a vector that is most similar to the residual transform coefficient among Nx1 vectors corresponding to indices of a code book of residual transform coefficients that are obtained by training of other pictures.
  • DCT discrete cosine transformation
  • the present invention also provides video data decoding method and an apparatus capable of reducing calculation complexity by storing in advance a table of residual coefficients obtained by performing inverse quantization and inverse DCT and indices matching the residual coefficients and looking up a residual coefficient corresponding to a received index from the table.
  • a video data encoding method comprising: generating a vector corresponding to a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; searching for a reference vector that is most similar to the vector among reference vectors corresponding to residual transform coefficient blocks of a sampled picture; and performing entropy encoding on an index matching the searched reference vector.
  • a video data encoding method comprising: generating a residual transform coefficient by transforming and quantizing a residual block that is a difference between a current block and a predicted block; and selecting a path among paths where entropy encoding is performed on the residual transform coefficient.
  • a video data encoding method comprising: generating vectors corresponding to residual coefficients; and clustering the vectors based on spatial nearness of the vectors.
  • a video data decoding method comprising: receiving vectors that correspond to residual transform coefficients obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block, wherein the vectors are allocated with indices; and performing entropy decoding on the vectors and storing the vectors with indices matching the vectors.
  • a video data decoding method comprising: extracting from a bitstream including index information an index matching a vector corresponding to a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; and reconstructing the residual block based on the extracted index.
  • a video data decoding method comprising: extracting path information from a bitstream, wherein the path information includes information on a first path where a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block is decoded and a second path where an index corresponding to the residual transform coefficient is decoded; and reconstructing the residual block that is the difference between the current block and the predicted block by performing entropy decoding, inversely quantizing, and inversely transforming the residual transform coefficient when the first path is selected.
  • a video data coding method comprising: generating a vector corresponding to a residual transform coefficient block obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; searching for a reference vector that is most similar to the vector among reference vectors corresponding to residual transform coefficient blocks of a sampled picture; and performing entropy encoding on an index matching the searched reference vector; and inversely searching for a vector corresponding to the index from the reference vectors; and performing inverse quantization and inverse transformation on the residual transform coefficient block reconstructed from the inversely-searched vector.
  • a video data coding method comprising: selecting a path among paths where entropy encoding is performed on a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; and encoding path information on the selected path; and decoding the encoded residual transform coefficient based on the path information.
  • an encoding apparatus comprising: a transformation quantization unit which generates a vector corresponding to a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; an optimal vector searching unit which searches for a reference vector that is most similar to the vector among reference vectors corresponding to residual transform coefficient blocks of a sampled picture; and an entropy encoding unit which performs entropy encoding on an index matching the searched reference vector.
  • an encoding apparatus comprising: a transformation quantization unit which generates a residual transform coefficient by transforming and quantizing a residual block that is a difference between a current block and a predicted block; and a path selection unit which selects a path among paths where entropy encoding is performed on the residual transform coefficient.
  • an encoding apparatus comprising: a vector generation unit which generates vectors corresponding to residual coefficients; and a clustering unit which performing clustering on the vectors based on spatial nearness of the vectors.
  • a decoding apparatus comprising: an entropy decoding unit which receives vectors that correspond to residual transform coefficients obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block and performs entropy decoding on the vectors, wherein the vectors are allocated with indices; and a storage unit which stores the vectors with indices matching the vectors.
  • a decoding apparatus comprising: an entropy decoding unit which extracts from a bitstream including index information an index matching a vector corresponding to a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; and a reconstructing unit which reconstructs the residual block based on the extracted index.
  • a coding apparatus a transformation quantization unit which generates a vector corresponding to a residual transform coefficient block obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; an optimal vector searching unit which searches for a reference vector that is most similar to the vector among reference vectors corresponding to residual transform coefficient blocks of a sampled picture; an entropy encoding unit which performs entropy encoding on an index matching the searched reference vector; an optimal vector inversely-searching unit which inversely searches for a vector corresponding to the index from the reference vectors; and an inverse-quantization inverse-transformation unit which performs inverse quantization and inverse transformation on the residual transform coefficient block reconstructed from the inversely-searched vector.
  • a coding apparatus a path selection unit which selects a path among paths where entropy encoding is performed on a residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between a current block and a predicted block; an entropy encoding unit which encodes path information on the selected path; and a decoding unit which decodes the encoded residual transform coefficient based on the path information.
  • a computer-readable medium having embodied thereon a computer program for executing the video data encoding and decoding methods and the coding methods according to the aforementioned aspects of the present invention.
  • FIG. 1 is a block diagram of a uni-path video data encoding apparatus according to an embodiment of the present invention.
  • FIG. 2 is a block diagram of a code book generation unit according to an embodiment of the present invention.
  • FIG. 3 is a view showing an operation of an optimal vector searching unit searching for a vector that is most similar to a residual transform coefficient from a code book according to an embodiment of the present invention.
  • FIG. 4 is a view showing an operation of reordering a code book in frames according to an embodiment of the present invention.
  • FIG. 5 is a block diagram of a dual-path video data encoding apparatus according to an embodiment of the present invention.
  • FIG. 6 is a block diagram of a uni-path video data decoding apparatus according to an embodiment of the present invention.
  • FIG. 7 is a block diagram of a dual-path video data decoding apparatus according to an embodiment of the present invention.
  • FIG. 8 is a flowchart for explaining a code book generating operation and a code book reordering operation according to an embodiment of the present invention.
  • FIG. 9 is a flowchart of a method of encoding a residual transform coefficient of video data according to an embodiment of the present invention.
  • FIG. 10 is a flowchart of a method of decoding a residual transform coefficient of video data according to an embodiment of the present invention.
  • FIG. 11 shows graphs of experimental results for video data encoding methods according to embodiments of the present invention.
  • FIG. 1 is a block diagram of a uni-path video data encoding apparatus according to an embodiment of the present invention.
  • the uni-path video data encoding apparatus 100 includes a motion estimation unit 102, a motion compensation unit 104, an intra-prediction unit 106, a subtraction unit 107, a transformation quantization unit 108, an optimal vector searching unit 110, an entropy encoding unit 112, an optimal vector inversely-searching unit 114, an inverse-quantization inverse-transformation unit 116, an addition unit 117, a frame memory 118, and a filter 120.
  • the motion estimation unit 102 performs motion estimation in order to search a region that is most similar to a current macroblock from a picture of a reference frame stored in the frame memory 118. More specifically, an area around the current macroblock in the reference frame is searched, and a region that is most similar to the current macroblock, that is, a region which has a minimum spatial difference from the current macroblock, is selected from the searched area. The spatial difference between the most similar region and the current block is output as a motion vector.
  • the motion compensation unit 104 reads the region that is most similar to the current macroblock from the picture of the reference frame stored in the frame memory
  • the motion estimation unit 102 and the motion compensation unit 104 cooperatively function as an inter-prediction unit which performs inter-prediction.
  • the inter-prediction may be performed on 16x8, 8x16, 8x8, 8x4, 4x8, 4x4 blocks as well as on a 16x16 macroblock.
  • the intra-prediction unit 106 performs intra-prediction by using correlation within the current frame.
  • the subtraction unit 107 calculates a difference between the current block and the predicted block to output a residual block that is a prediction difference block.
  • the transformation quantization unit (sometimes, referred to as a quantization transformation unit) 108 performs discrete cosine transformation (DCT) and quantization of the prediction difference block obtained by performing the inter-prediction in order to output a residual transform coefficient block and then reorders the residual transform coefficient block into a 1-D array by using zigzag scan and Run-Level encoding.
  • the optimal vector searching unit 110 searches for a vector that is most similar to a vector that is a 1-D array of the residual transform coefficient block obtained by performing the transformation and quantization in a code book, and detects an index matching the most similar vector. The aforementioned operations will be described later in detail with reference to FIG. 3.
  • the entropy encoding unit 112 encodes the index matching the most-similar vector by using a variable length coding (VLC) scheme to output a final bitstream.
  • VLC variable length coding
  • the optimal vector inversely-searching unit 114 searches for the vector corresponding to the matched index from the code book in an inverse manner where the operations of the optimal vector searching unit 110 are inversely performed to select the vector.
  • the inverse-quantization inverse-transformation unit 116 (sometimes referred to as an inverse quantization transformation unit) performs inverse quantization and inverse transformation on the selected vector so as to output a residual block that is the prediction difference block.
  • the addition unit 117 adds the residual block to the inter-predicted block or the intra-predicted block. Finally, the picture is reconstructed by the filter 120.
  • FIG. 2 is a block diagram of a code book generation unit according to an embodiment of the present invention.
  • a code book is generated by using vector quantization.
  • the code book generation unit 200 includes a vector generation unit 210, a clustering unit 220, an index allocation unit 230, and a storage unit 240.
  • the vector generation unit 210 generates vectors corresponding to residual coefficients. For example, a large number of residual transform coefficients in units of a 4x4 block, which are obtained by DCT and quantization of real sample pictures are reordered as a 16x1 vector in a space.
  • the residual coefficient may be any one of a residual transform coefficient obtained by performing motion compensated transformation, a residual transform coefficient obtained by transforming a residual block that is a difference between a current block and a predicted block, and a residual transform coefficient obtained by transforming and quantizing the residual block.
  • a residual transform coefficient obtained by transforming a residual block that is a difference between a current block and a predicted block and a residual transform coefficient obtained by transforming and quantizing the residual block.
  • an embodiment of the present invention is described by using the residual transform coefficient obtained by transforming the residual block that is a difference between the current block and the predicted block.
  • the same description of the present embodiment can be applied to other residual coefficients.
  • the clustering unit 220 performs clustering of the residual transform coefficients by using a clustering analysis algorithm.
  • the clustering is an operation of generating a code book by clustering objects near to each other based on positions thereof in a 16-D space of the reordered vectors.
  • a K-means algorithm may be used as an example of the clustering analysis algorithm.
  • the clustering includes clustering of residual coefficients (residual data) obtained by a motion compensated (MC) transform coder, clustering of quantized residual signals of transformed signals, and clustering of transformed residual signals.
  • MC motion compensated
  • the residual transform coefficients greatly depend on a quantization parameter.
  • the residual transform coefficients may be distributed in units of a clustering number that suits the quantization parameter. For example, a range of the quantization parameter is partitioned into predetermined regions, and the residual transform coefficients may be clustered in units of a clustering number that suits each of the predetermined regions.
  • a code book of the clustered residual transform coefficients is an M x N table in which M represents the number of to-be-allocated indices and N represents a dimension of residual block vectors.
  • the storage unit 240 stores the code book as an initial code book. When the vectors of the code book are reordered later, the storage unit 240 stores the code book of the reordered vectors.
  • the index allocation unit 230 allocates the indices and Bin strings that are binarized values of the indices to the vectors of the code book.
  • the binarization of the indices is needed so as to encode the indices in Context Adaptive Binary Arithmetic Coding (CABAC) that is an entropy encoding scheme.
  • CABAC Context Adaptive Binary Arithmetic Coding
  • an input symbol is represented by the Bin string, that is, a combination of 0 and 1.
  • the symbol can be encoded with a smaller number of bits. Therefore, in order to efficiently binarize the indices, the vectors after the clustering may be reordered in a descending order of a clustering density.
  • a vector having a higher clustering density can have a shorter Bin string.
  • unary 3rd order exponential Golomb coding may be used.
  • unary is used for the four upper vectors.
  • Table 1 is an example of a code book representing clustered 16x1 vectors, indices allocated to the vectors, and Bin strings obtained by the aforementioned binarization of the indices.
  • there are a large number of residual transform coefficients of 0. Therefore, by using 1-bit flags, a vector having all the residual transform coefficients of 0 is distinguished from other vectors, so that encoding efficiency can be increased.
  • FIG. 3 is a view showing an operation of an optimal vector searching unit searching for a vector that is most similar to a residual transform coefficient from a code book according to an embodiment of the present invention.
  • Equation 1 a Euclean distance equation represented by Equation 1.
  • Equation 1 All the distances of the residual transform coefficient from all the vectors of the code book are calculated by using Equation 1 , and a vector having the shortest distance is determined as an optimal vector, that is, the most-similar vector. If a distance from a vector is calculated to be 0, the distance calculation for subsequent vectors does not proceed, and the vector is determined as the optimal vector.
  • C denotes an arbitrary 16x1 vector of the code book
  • R denotes a residual transform coefficient
  • Distance denotes a Euclean distance between a residual transform coefficient and an arbitrary vector of the code book.
  • the index corresponding to the vector is encoded by the entropy encoding unit.
  • FIG. 4 is a view showing an operation of reordering a code book in' frames (hereinafter, referred to as a frame-by-frame code book reordering operation) according to an embodiment of the present invention.
  • the vectors of the code book are reordered in a descending order of the selection times. Therefore, a smaller number of bits can be used to adaptively encode the indices according to time and picture types. Namely, as the value of an index becomes larger, the length of the corresponding Bin string becomes longer.
  • a distribution of probabilities of indices to the previous frame is calculated, and by using the probabilities of the indices, the vectors of the code book are reordered in the descending order of the probabilities of the indices.
  • the indices are reallocated from Index 0 at the uppermost of the code book.
  • the aforementioned code book reordering operation for the encoder can be performed on a decoder storing the same code book. Accordingly, any additional parameters need to be transmitted.
  • the frame-by-frame code book reordering also needs to be performed on the decoder, so that calculation complexity increases.
  • the reordering is only performed on about 20 upper indices. Accordingly, it is possible to improve performance of reordering in comparison with the existing reordering scheme.
  • a code book 410 is a before-reordering code book.
  • the code book 410 shows the selection times 430 of the vectors corresponding to the indices used for one frame encoding.
  • selection Times denotes the selection times of the vectors used in the code book 410.
  • a code book 420 is an after-reordering code book which is reordered according to the selection times 430. If the vector of Index 3 having large selection times (25) and the vector of Index 2 having small selection times (11 ) are reordered so as to exchange positions with each other in the code book 410, the vector of Index 3 having the larger selection times has a shorter Bin string in the code book 420. In particular, since the vector having all the residual transform coefficients of 0 is used in all the frames by the largest selection times, the vector does not need to be reordered, and the vector can be always positioned at Index 0 in a code book (not shown).
  • FIG. 5 is a block diagram of a dual-path video data encoding apparatus according to an embodiment of the present invention.
  • the dual-path video data encoding apparatus 500 has a construction similar to that of the aforementioned uni-path video data encoding apparatus 100. However, in the dual-path video data encoding apparatus 500, the residual transform coefficient block generated by DCT and quantization of the transformation quantization unit 510 is arranged to proceed to two paths (Paths 1 and 2) at a branch point.
  • the dual-path video data encoding apparatus 500 includes a transformation quantization unit 510, an optimal vector searching unit 520, an entropy encoding unit 530, an optimal vector inversely-searching unit 540, an inverse-quantization inverse-transformation unit 550, an addition unit 560, and a path selection unit 570.
  • the path selection unit 570 selects a path (Path 1 or 2) to allow the residual transform coefficient to proceed to the corresponding paths and to perform an entropy encoding.
  • Path 1 corresponds to the operation of the aforementioned uni-path video data encoding apparatus 100.
  • the optimal vector searching unit 520 searches for a vector that is most similar to a 16x1 vector corresponding to a 1-D array of the residual transform coefficient blocks in the code book, and the entropy encoding unit
  • Path 2 corresponds to an existing standard compression operation.
  • Path 2 without selection and transformation of indices performed based on the code book by the optimal vector searching unit 520, the residual transform coefficient blocks are input to the entropy encoding unit 530.
  • a scheme for comparing a rate-distortion cost (RD cost) of Path 1 with the RD cost of Path 2 and for selecting a path having a smaller RD cost is used as a path selection scheme for the encoding of the residual transform coefficients of the current macroblock.
  • a flag bit (one bit per macroblock) which is inserted into the bitstream to represent a path may be transmitted to the decoder to identify the used path. Accordingly, the decoding can be performed according to the used path.
  • the RD cost can be calculated by using Equation 2 represented by a Lagrangian cost function.
  • Equation 2 “Rates” denotes the amount of bits used for encoding the residual transform coefficient
  • “Distortion” denotes a degree of distortion of a reconstructed macroblock with respect to an original macroblock, in which the degree of distortion is calculated based on a sum of square difference (SSD).
  • B(k,l) denotes a value of a pixel (k,l) of the original macroblock
  • B'(k,l) denotes a value of a pixel (k,l) of the reconstructed macroblock.
  • ⁇ is a constant determined according to a quantization parameter.
  • the index of the vector encoded through Path 1 is input to the optimal vector inversely-searching unit 540 which searches for the vector corresponding to the index.
  • the corresponding vector is subjected to inverse-quantization and inverse-transformation operations in the inverse-quantization inverse-transformation unit 550 and, subsequently, added to the inter-predicted block in the addition unit 560, so that a picture can be reconstructed.
  • the residual transform coefficient encoded through Path 2 is subjected to the inverse-quantization and inverse-transformation operations in the inverse-quantization inverse-transformation unit 550, and subsequently, added to the inter-predicted block in the addition unit 560, so that a picture can be reconstructed.
  • FIG. 6 is a block diagram of a uni-path video data decoding apparatus according to an embodiment of the present invention.
  • the uni-path video data decoding apparatus 600 includes an entropy decoding unit 610, an optimal vector inversely-searching unit 620, an inverse-quantization inverse-transformation unit 630, an addition unit 640, an intra-prediction unit 650, and a motion compensation unit 660.
  • the entropy decoding unit 610 When the entropy decoding unit 610 receives a bitstream including an index matching a vector corresponding to the residual transform coefficient obtained by transforming and quantizing a residual block that is a difference between the current block and the predicted block, the entropy decoding unit 610 performs entropy decoding on the bitstream in order to output the index.
  • the optimal vector inversely-searching unit 620 searches for a vector corresponding to the index output from the entropy decoding unit 610 from a code book that is stored in the decoding apparatus in advance, so that the residual transform coefficient block can be reconstructed from the vector.
  • the code book of the decoder is obtained by receiving from the encoder (or encoding apparatus) an M x N table of clustered vectors of the residual transform coefficients of a large number of sampled pictures and decoding and storing the table.
  • M denotes the number of indices
  • N denotes a dimension of residual transform coefficients.
  • a table of residual coefficients may be generated by performing inverse quantization and inverse transformation on the residual transform coefficients matching the indices of the code book in advance. Accordingly, without the inverse quantization and the inverse transformation of the residual transform coefficients corresponding to the received indices, the residual coefficients can be obtained by directly looking up the table.
  • probabilities of indices received in frames may be calculated according to the selection times of the vectors corresponding to the indices.
  • the vectors of the code book may be reordered in a descending order of the probabilities, and the indices are reallocated to the vectors.
  • the inverse-quantization inverse-transformation unit 630 performs the inverse quantization and the inverse transformation on the reconstructed residual transform coefficient blocks in order to generate residual coefficient blocks.
  • the motion compensation unit 660 generates predicted blocks.
  • the predicted blocks are added to the residual coefficient blocks in the addition unit 640, so that the picture can be reconstructed.
  • FIG. 7 is a block diagram of a dual-path video data decoding apparatus according to an embodiment of the present invention. For simplicity, a detailed description of the components that are the same as those of the aforementioned uni-path video data decoding apparatus 600 is omitted.
  • the dual-path video data decoding apparatus 700 includes an entropy decoding unit 710, an optimal vector inversely-searching unit 720, an inverse-quantization inverse-transformation unit 730, and an addition unit 760.
  • the entropy decoding unit 710 extracts path information from a received bitstream.
  • a flag bit one bit, which is inserted into the bitstream to represent a path, is decoded.
  • the path (Path 1 or 2) is selectively changed according to the flag bit.
  • the entropy decoding unit 710 When Path 1 is selected, the entropy decoding unit 710 performs entropy decoding on the bitstream in an decoding scheme that is the same as that of the uni-path video data decoding apparatus 600 in order to extract the index. When Path 2 is selected, the residual transform coefficient blocks are subjected to the entropy decoding.
  • the optimal vector inversely-searching unit 720 searches for the vector corresponding to the extracted index from the code book and selects the vector.
  • the inverse-quantization inverse-transformation unit 730 performs the inverse quantization and the inverse transformation on the residual transform coefficient block corresponding to the selected vector and the residual transform coefficient block obtained through Path 2 in order to generate a residual coefficient block.
  • the motion compensation unit 780 generates predicted blocks.
  • the predicted blocks are added to the residual coefficient blocks output through Path 1 or 2 in the addition unit 760, so that the picture can be reconstructed.
  • FIG. 8 is a flowchart for explaining a code book generating operation and a code book reordering operation according to an embodiment of the present invention.
  • the residual coefficient may be any one of a residual transform coefficient obtained by performing motion compensated transformation on a sampled picture, a residual transform coefficient obtained by transforming a residual block that is a difference between a current block and a predicted block, and a residual transform coefficient obtained by transforming and quantizing the residual block.
  • the code book which is obtained by clustering the vectors based on spatial nearness is stored (S820).
  • the code book is an M x N table of vectors, in which "M" and “N” denote the number of indices and a dimension of residual transform coefficients, respectively.
  • the indices and Bin strings that are binarized values of the indices are allocated to the clustered vectors (S830).
  • an initial code book allocated initial indices and Bin strings, the selection times of the vectors corresponding to the indices selected in frames of an input video are calculated.
  • the vectors of the code book are reordered according to a distribution of probabilities of indices per frame in a descending order of probabilities of indices, and the indices are also reallocated (S840).
  • FIG. 9 is a flowchart of a method of encoding a residual transform coefficient of video data according to an embodiment of the present invention.
  • a residual block that is a difference between a current block and an predicted block of input video data is subjected to transformation and quantization in a transformation quantization unit in order to generate a vector that is a 1-D array of the residual transform coefficient block (S910).
  • An optimal vector searching unit searches for an optimal vector that is most similar to the vector from a code book that is a table of reference vectors which are stored in advance and selects the optimal vector (S920).
  • an entropy encoding unit performs entropy encoding on an index matching the selected optimal vector to generate a final bitstream (S930).
  • a path selection operation may be provided.
  • the residual transform coefficient blocks can be selectively subjected to the entropy encoding similarly to an existing method (S940).
  • FIG. 10 is a flowchart of a method of decoding a residual transform coefficient of video data according to an embodiment of the present invention.
  • an entropy decoding unit performs entropy decoding on a received bitstream where indices are encoded by using a code book (S1010).
  • a flag bit which is inserted into the bitstream to represent a path, is decoded, and the paths are selectively changed according to the flag bit (S1020).
  • an optimal vector corresponding to the index is searched for from a code book including decoded indices in advance and which is the same as a code book of an encoder in an inverse manner to output a residual transform coefficient block (S1030).
  • the residual transform coefficient block is decoded by the entropy decoding unit, and subsequently, an inverse quantization operation is performed.
  • the decoded residual transform coefficient block is subjected to the inverse quantization and the inverse transformation in the inverse quantization inverse transformation unit in order to generate a residual coefficient block (S1040).
  • the motion compensation unit performs inter prediction by using the video data included in the bitstream to generate a predicted block (S1050).
  • FIG 11 shows graphs of experimental results for video data encoding methods according to embodiments of the present invention.
  • Graph (a) shows an R-D curve of a Foreman picture
  • Graph (b) shows an R-D curve of a Mobile picture.
  • Each of the R-D curve (a) of the Foreman picture and the R-D curve (b) of the Mobile pictures shows experimental results of performance of video data encoding methods (uni-path and dual-path) according to embodiments of the present invention and encoding methods using H.264 CABAC and H.264 context- adaptive variable length coding (CAVLC) of joint model (JM) 8.6, that is, an H.264 reference encoder.
  • H.264 CABAC and H.264 context- adaptive variable length coding (CAVLC) of joint model (JM) 8.6 that is, an H.264 reference encoder.
  • variable length motion estimation In addition, variable length motion estimation, rate-distortion optimization, IPPP structure, 30-frame-period intra frame, 5 reference frames, and ⁇ 16 motion vector searching regions are used.
  • the video data encoding methods according to the embodiments of the present invention shows a high performance of 0.3dB to 0.5dB, which is substantially equal to performance of the H.264 CABAC.
  • the invention can also be embodied as computer readable codes on a computer readable recording medium.
  • the computer readable recording medium is any data storage device that can store data which can be thereafter read by a computer system. Examples of the computer readable recording medium include read-only memory (ROM), random-access memory (RAM), CD-ROMs, magnetic tapes, floppy disks, optical data storage devices, and carrier waves (such as data transmission through the Internet).
  • ROM read-only memory
  • RAM random-access memory
  • CD-ROMs compact discs
  • magnetic tapes magnetic tapes
  • floppy disks optical data storage devices
  • carrier waves such as data transmission through the Internet
  • carrier waves such as data transmission through the Internet
  • the computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion. Also, functional programs, codes, and code segments for accomplishing the present invention can be easily construed by programmers skilled in the art to which the present invention pertains.
  • indices corresponding to residual transform coefficients obtained by performing DCT and quantization are encoded and transmitted, so that it is possible to increase compression ratio.
  • a table of residual coefficients obtained by performing inverse quantization and inverse DCT (or other inverse transformation) is stored, and a residual coefficient corresponding to an index included in a received bitstream is searched for by looking up the table, so that inverse DCT and inverse quantization operations can be omitted. Accordingly, it is possible to reduce calculation complexity.

Abstract

L'invention concerne des procédés et des appareils de codage et de décodage de données vidéo. En ce qui concerne les procédés de codage et de décodage de données vidéo, des tables de codage sont fournies à un codeur et à un décodeur. Concernant le codeur, un indice correspondant à un vecteur particulièrement similaire à un vecteur courant d'une image mobile entrée parmi les vecteurs de la table de codage est codé. Concernant le décodeur, l'indice est décodé. En conséquence, on peut augmenter le taux de compression et réduire la complexité des calculs.
PCT/KR2007/000082 2006-01-07 2007-01-05 Procédé et appareil de codage et de décodage de données vidéo WO2007081116A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US12/160,154 US20090067503A1 (en) 2006-01-07 2007-01-05 Method and apparatus for video data encoding and decoding

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
KR20060002042 2006-01-07
KR10-2006-0002042 2006-01-07
KR1020070001386A KR100867995B1 (ko) 2006-01-07 2007-01-05 동영상 데이터 인코딩/디코딩 방법 및 그 장치
KR10-2007-0001386 2007-01-05

Publications (1)

Publication Number Publication Date
WO2007081116A1 true WO2007081116A1 (fr) 2007-07-19

Family

ID=38256494

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2007/000082 WO2007081116A1 (fr) 2006-01-07 2007-01-05 Procédé et appareil de codage et de décodage de données vidéo

Country Status (1)

Country Link
WO (1) WO2007081116A1 (fr)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009038422A3 (fr) * 2007-09-20 2009-05-22 Lg Electronics Inc Procédé et appareil de traitement de signal
CN102714720A (zh) * 2009-11-18 2012-10-03 Sk电信有限公司 通过选择一组预测候选运动向量来对运动向量进行编码/解码的方法和装置以及使用所述方法和装置进行图像编码/解码的方法和装置
CN109120927A (zh) * 2011-11-04 2019-01-01 夏普株式会社 图像解码装置、图像解码方法以及图像编码装置
CN110830799A (zh) * 2012-01-18 2020-02-21 韩国电子通信研究院 视频解码装置、视频编码装置和计算机可读记录介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5768438A (en) * 1994-10-19 1998-06-16 Matsushita Electric Industrial Co., Ltd. Image encoding/decoding device
US5861923A (en) * 1996-04-23 1999-01-19 Deawoo Electronics Co., Ltd. Video signal encoding method and apparatus based on adaptive quantization technique
US5905812A (en) * 1996-05-02 1999-05-18 Samsung Electronics Co., Ltd. Perceptual error processing method and image coding device using the same

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5768438A (en) * 1994-10-19 1998-06-16 Matsushita Electric Industrial Co., Ltd. Image encoding/decoding device
US5861923A (en) * 1996-04-23 1999-01-19 Deawoo Electronics Co., Ltd. Video signal encoding method and apparatus based on adaptive quantization technique
US5905812A (en) * 1996-05-02 1999-05-18 Samsung Electronics Co., Ltd. Perceptual error processing method and image coding device using the same

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9031851B2 (en) 2007-09-20 2015-05-12 Lg Electronics Inc. Method and an apparatus for processing a signal
WO2009038422A3 (fr) * 2007-09-20 2009-05-22 Lg Electronics Inc Procédé et appareil de traitement de signal
US9479793B2 (en) 2009-11-18 2016-10-25 Sk Telecom Co., Ltd. Method and apparatus for encoding/decoding a motion vector by selecting a set of predicted candidate motion vectors, and method and apparatus for image encoding/decoding using the same
CN105072451B (zh) * 2009-11-18 2018-05-29 Sk电信有限公司 视频解码装置
CN104954802A (zh) * 2009-11-18 2015-09-30 Sk电信有限公司 视频编码方法
CN105072450A (zh) * 2009-11-18 2015-11-18 Sk电信有限公司 对运动向量编码的装置以及对运动向量解码的装置和方法
CN105072451A (zh) * 2009-11-18 2015-11-18 Sk电信有限公司 视频解码装置
US9363530B2 (en) 2009-11-18 2016-06-07 Sk Telecom Co., Ltd. Method and apparatus for encoding/decoding a motion vector by selecting a set of predicted candidate motion vectors, and method and apparatus for image encoding/decoding using the same
CN102714720A (zh) * 2009-11-18 2012-10-03 Sk电信有限公司 通过选择一组预测候选运动向量来对运动向量进行编码/解码的方法和装置以及使用所述方法和装置进行图像编码/解码的方法和装置
CN104954803A (zh) * 2009-11-18 2015-09-30 Sk电信有限公司 视频编码装置
CN105072450B (zh) * 2009-11-18 2018-06-12 Sk电信有限公司 对运动向量编码的装置以及对运动向量解码的装置和方法
CN104954802B (zh) * 2009-11-18 2018-09-14 Sk电信有限公司 视频编码方法
CN109120927A (zh) * 2011-11-04 2019-01-01 夏普株式会社 图像解码装置、图像解码方法以及图像编码装置
CN109120927B (zh) * 2011-11-04 2021-05-18 夏普株式会社 图像解码装置、图像解码方法以及图像编码装置
CN110830799A (zh) * 2012-01-18 2020-02-21 韩国电子通信研究院 视频解码装置、视频编码装置和计算机可读记录介质
US11706438B2 (en) 2012-01-18 2023-07-18 Electronics And Telecommunications Research Institute Method and device for encoding and decoding image
CN110830799B (zh) * 2012-01-18 2023-09-15 韩国电子通信研究院 视频解码装置、视频编码装置和传输比特流的方法

Similar Documents

Publication Publication Date Title
US20090067503A1 (en) Method and apparatus for video data encoding and decoding
US10257517B2 (en) Method and apparatus for entropy encoding using hierarchical data unit, and method and apparatus for decoding
CN111819853B (zh) 图像块编码装置和图像块编码方法
CN108235023B (zh) 用于编码和解码图像的方法、编码和解码设备
US8625670B2 (en) Method and apparatus for encoding and decoding image
CN107396126B (zh) 用于编码和解码图像的方法、编码和解码设备、以及相应的计算机程序
US6925119B2 (en) Process and system for the compression of digital video signals, a system and a computer program product therefor
KR101108681B1 (ko) 동영상 코덱에서의 주파수 변환 계수 예측 방법 및 장치,이를 구비한 부호화 및 복호화 장치와 방법
KR101611437B1 (ko) 복수의 프레임을 참조하여 영상을 부호화, 복호화하는 방법 및 장치
JP7110441B2 (ja) デジタル画像のコーディング方法、デコーディング方法、装置、および関連するコンピュータプログラム
US20090225843A1 (en) Method and apparatus for encoding and decoding image
CN104221373A (zh) 用于采样自适应偏移代码化和/或信号通知的设备和方法
EP2767086A1 (fr) Transformée la plus probable pour codage à prédiction intra
US20170134732A1 (en) Systems and methods for digital media communication using syntax planes in hierarchical trees
US10771815B2 (en) Method and apparatus for processing video signals using coefficient induced prediction
WO2007081116A1 (fr) Procédé et appareil de codage et de décodage de données vidéo
CN114175635A (zh) 用于推导帧内预测模式的方法及装置
CN113170209A (zh) 图像编码/解码方法和设备以及存储比特流的记录介质
US20220353505A1 (en) Method for reconstructing residual blocks of chroma blocks, and video decoding apparatus
KR101796876B1 (ko) 움직임 추정을 이용한 영상 부호화 방법 및 장치
KR101662741B1 (ko) 영상 복호화 방법
KR101249346B1 (ko) 적응적 양자화 계수 탐색을 이용한 영상 부호화/복호화 방법 및 장치, 상기 방법을 기록한 컴퓨터로 판독 가능한 기록매체
US20200029079A1 (en) Method for processing image providing improved arithmetic encoding, method for decoding and encoding image using same, and apparatus for same
KR101525015B1 (ko) 영상의 부호화 방법 및 장치, 영상 복호화 방법 및 장치
KR101618214B1 (ko) 영상 복호화 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application
WWE Wipo information: entry into national phase

Ref document number: 12160154

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 07708434

Country of ref document: EP

Kind code of ref document: A1