WO2013064109A1 - 一种图像编码、解码的方法和装置 - Google Patents

一种图像编码、解码的方法和装置 Download PDF

Info

Publication number
WO2013064109A1
WO2013064109A1 PCT/CN2012/084062 CN2012084062W WO2013064109A1 WO 2013064109 A1 WO2013064109 A1 WO 2013064109A1 CN 2012084062 W CN2012084062 W CN 2012084062W WO 2013064109 A1 WO2013064109 A1 WO 2013064109A1
Authority
WO
WIPO (PCT)
Prior art keywords
quantization matrix
quantization
matrix
image
encoding
Prior art date
Application number
PCT/CN2012/084062
Other languages
English (en)
French (fr)
Inventor
杨海涛
周建同
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP12846329.6A priority Critical patent/EP2750386A4/en
Publication of WO2013064109A1 publication Critical patent/WO2013064109A1/zh
Priority to US14/268,115 priority patent/US9667958B2/en
Priority to US15/492,868 priority patent/US10091531B2/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods 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
    • H04N19/124Quantisation
    • H04N19/126Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/136Incoming video signal characteristics or properties
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169Methods 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
    • H04N19/17Methods 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
    • H04N19/176Methods 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
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/46Embedding additional information in the video signal during the compression process
    • H04N19/463Embedding additional information in the video signal during the compression process by compressing encoding parameters before transmission
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90Methods 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
    • H04N19/91Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Definitions

  • the present invention relates to the field of communications, and in particular to a method and apparatus for image encoding and decoding. Background technique
  • video data Due to the huge amount of video data, in practical applications, video data usually requires compression coding processing.
  • the encoder processes the video data through a prediction, transform, quantization, and entropy encoding process to effect data compression to generate a video stream.
  • the video stream can be used for storage or network transmission.
  • the decoder performs a decoding operation on the video code stream by entropy decoding, inverse quantization, inverse transform, and prediction compensation to reconstruct the video data.
  • the H.264 encoding technique achieves precise control of signal compression distortion by using a quantization matrix (quant i ta t ive ma t ix, QM).
  • QM quantization matrix
  • the encoder gives a set of QMs suitable for the current encoded image based on the current image content, and then writes the QM encoding to the codestream.
  • the decoder After receiving the code stream with the QM information, the decoder decodes the QM information and uses the QM information to decode the image.
  • there can be up to eight sets of QM matrices QMi, i l, 2, ... 8 per frame.
  • the 8 groups of QM matrices respectively represent the luminance ⁇ , chrominance Cb, chrominance Cr, and inter-predicted luminance Y, chrominance Cb, and chrominance Cr in the 4*4 transform, and 8* in the Q*; 8 transform, intra-prediction and inter-prediction luminance Y two QM. Since the amount of QM information is large, the QM information needs to be compression-coded to reduce the number of bits used to represent the QM information.
  • H. 264 uses the following compression method to encode six 4*4 quantization matrices and two two 8 *8 quantization matrices, respectively. The specific steps are as follows:
  • the first step is to perform a scanning operation on the two-dimensional quantization matrix to generate one-dimensional data
  • the second step is to perform DPCM encoding on the one-dimensional data
  • the encoded data is entropy encoded and written into the code stream.
  • the transform and quantization used in the above scheme are both N*N square matrices.
  • N*M quantization matrix When non-square transform and quantization matrix are used, for the N*M quantization matrix, according to the above quantization matrix compression method, To many bits represent the quantization matrix, when the application bandwidth is small, the bits used to transmit the quantization matrix will seriously affect the quality of the encoded image.
  • Embodiments of the present invention provide methods and apparatus for image encoding and decoding to reduce the transmission bandwidth of a code stream.
  • An embodiment of the present invention provides an image encoding method, where the encoding method includes: performing predictive encoding on an image; performing transform encoding on the image subjected to predictive encoding; and quantizing the transform encoded image using a quantization matrix.
  • the quantization matrix is a matrix reflecting image quantization step size information
  • the quantization matrix includes an M*N quantization matrix, an N*M quantization matrix, and the N*M quantization matrix is rotated by the M*N quantization matrix
  • An embodiment of the present invention provides a method for decoding an image, where the decoding method includes: performing entropy decoding on a received code stream to obtain image data and a quantization matrix, where the quantization matrix is a matrix reflecting image quantization step size information.
  • the quantization matrix includes an M*N quantization matrix; an N*M quantization matrix is obtained by transposition from the M*N quantization matrix; using the M*N quantization matrix, the N*M quantization matrix pair to be entropy decoded
  • the image data is inverse quantized; the inversely quantized image data is inversely transformed; the inversely transformed image data is subjected to prediction compensation to generate a decoded image.
  • An embodiment of the present invention provides an image encoding apparatus, where the encoding apparatus includes: a prediction encoding module, configured to perform predictive encoding on an image; and a transform encoding module, configured to transform and encode the image that is subjected to predictive encoding; And an encoding module, configured to perform quantization coding on the transform-coded image by using a quantization matrix, where the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes an M*N quantization matrix and an N*M quantization matrix.
  • the N*M quantization matrix is obtained by transposition of the M*N quantization matrix; an entropy coding module, configured to entropy encode the quantized image, and encode the M*N quantization matrix to generate Code stream.
  • the embodiment of the present invention provides an image decoding apparatus, where the decoding transposition includes: an entropy decoding unit, configured to perform entropy decoding on the received code stream to obtain image data and a quantization matrix, where the quantization matrix is a reflection image.
  • the quantization matrix includes an M*N quantization matrix; an inverse quantization unit, configured to obtain an N*M quantization matrix by transposition from the M*N quantization matrix, using the M*N quantization matrix And the N*M quantization matrix performs inverse quantization on the entropy-decoded image data; the inverse transform unit is configured to inverse-transform the inverse-quantized image data; and the prediction compensation unit is configured to perform inverse-transformed image data. Predictive compensation, generating a decoded image.
  • the embodiment of the invention effectively saves the number of bits required to encode the quantization matrix and improves the compression efficiency.
  • FIG. 1 is a flow chart of an embodiment of an image encoding method of the present invention
  • FIG. 2 is a flow chart of an embodiment of an image decoding method of the present invention.
  • FIG. 3 is a flow chart of still another embodiment of an image encoding method of the present invention.
  • FIG. 4 is a flow chart of still another embodiment of an image decoding method according to the present invention.
  • FIG. 5 is a flowchart of still another embodiment of an image encoding method according to the present invention.
  • FIG. 6 is a flowchart of still another embodiment of an image decoding method according to the present invention.
  • Figure 7 is a structural diagram of an embodiment of an image coding apparatus according to the present invention.
  • FIG. 8 is a block diagram showing an embodiment of an image decoding apparatus of the present invention. detailed description The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is a partial embodiment of the invention, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.
  • Embodiments of the present invention provide an image encoding method.
  • Figure 1 provides a flow chart of one embodiment of the method.
  • the encoding method includes:
  • S 1 05 performs quantization coding on the transform-coded image using a quantization matrix, where the quantization matrix is a matrix reflecting image quantization step size information, and the quantization matrix includes an M*N quantization matrix and an N*M quantization matrix.
  • the N*M quantization matrix is obtained by transposition of the M*N quantization matrix;
  • each frame image is segmented into smaller image blocks for encoding or decoding processing, such as dividing one frame image into image blocks of size NxN.
  • S 1 01 specifically includes: when encoding an N ⁇ N image block, the NxN image is predictively encoded according to different sub-blocks. For flat image areas, use larger image sub-block partitioning to save description information for describing the image partitioning relationship; for detailed image areas, use smaller image sub-blocking to improve image encoding. Predictive accuracy; improve compression efficiency through adaptive partitioning.
  • the predictive coding includes intra coding and inter coding depending on the prediction relationship. Intra coding uses spatially adjacent reconstructed pixels from the same frame of pictures to predict the current coded block. Inter prediction predicts the current coded block using the pixels of the picture at the time before or after the picture at which the current coded block is located.
  • S 1 03 specifically includes: transform coding by transform coding the predictively encoded data to concentrate data energy and reduce bits used to describe the data. Quantity.
  • Transform coding techniques include DCT, DST, wavelet transform, and the like.
  • transform coding is performed using different transform matrices. Taking the DCT transform as an example, for a 4x4 sub-block, a 4x4 DCT transform is performed, an 8x8 sub-block performs an 8x8 DCT transform, a 32x32 sub-block roast performs a 32x32 DCT transform, and a 4x4 or 8x8 transform can also be performed; or Non-square transform, 32x8, 16x4, etc.
  • the advantage of non-square transformation is that the image content can be better adapted and the coding efficiency is improved.
  • S105 specifically includes: when encoding a video signal, selecting a transform matrix suitable for the image content according to different image content, and performing quantization coding on the transform-coded image by using a quantization matrix, At the same time, the image data information and the transform matrix are encoded into the code stream to control the compression efficiency of the video data.
  • the quantization matrix includes an M*N quantization matrix and an N*M quantization matrix, and the N*M quantization matrix is obtained by transposition from the M*N quantization matrix.
  • the N*M quantization matrix is a transposed matrix of the M*N quantization matrix.
  • QM1-QM6 are the quantization matrices of Y, Cb and Cr components for 4*4 transform matrix interframe and intra prediction coding, respectively;
  • QM7_QM10 are 8*4 and 4*8 transform matrix frames respectively The quantization matrix of the Y component when inter- and intra-predictive coding;
  • QM11-QM12 are quantization matrices of the Y component of the 8*8 transform matrix interframe and intra prediction coding, respectively.
  • QM7 and QM9 are transposed, and QM8 and QM10 are transposed.
  • the current image transformed image is quantized and encoded by the quantization matrix, and written into the code stream.
  • the quantization matrix is encoded.
  • QM9 and 10 are not encoded.
  • the information of QM9 and 10 is obtained by transposing the information of QM7 and 8 through the decoding end.
  • the N*M quantization matrix is obtained by transposition by the M*N quantization matrix, including: transposing according to the N*M quantization matrix and the M*N quantization matrix
  • the difference value of the matrix is calculated to obtain an N*M difference quantization matrix
  • the encoding of the M*N quantization matrix includes: performing the quantized and encoded image, the M*N quantization matrix, and the M*N difference quantization matrix Entropy coding.
  • S107 specifically includes: encoding, for each quantization matrix, a per-port method:
  • the predicted difference data is entropy encoded and written into the code stream.
  • S107 specifically includes: encoding, for each quantization matrix, the following method:
  • the prediction signal is derived from the signal after transposing the QM NxM , and the predicted quantization matrix difference signal DQM MxN is obtained ;
  • the one-dimensional coefficients are entropy encoded and written into the code stream.
  • Embodiments of the present invention provide a method for decoding an image.
  • Figure 2 provides a flow chart of one embodiment of the method.
  • the decoding method includes:
  • S201 performs entropy decoding on the received code stream to obtain image data and a quantization matrix, where the quantization matrix is a matrix reflecting image quantization step size information, and the quantization matrix includes an M*N quantization matrix;
  • S 203 obtains an N*M quantization matrix by transposition from the M*N quantization matrix
  • S205 performing inverse quantization on the entropy decoded image data by using the M*N quantization matrix and the N*M quantization matrix;
  • S209 performs prediction compensation on the inversely transformed image data to generate a decoded image.
  • the process of entropy decoding the received code stream by S201 to obtain a quantization matrix includes:
  • the entropy decoding quantizes the matrix code stream to obtain a one-dimensionally predicted quantized coefficient difference signal; and performs DPCM prediction compensation on the one-dimensional quantized coefficient difference signal;
  • the process of entropy decoding the received code stream by S201 to obtain a quantization matrix includes:
  • the transposed signal of the QM NxM signal that has been decoded is used as a predicted value, and the predicted value is used for predictive compensation of 0 ( ⁇ to obtain a reconstructed QM MxN signal, and the decoding process of the QM MxN is completed.
  • S203 obtains an N*M quantization matrix by transposition from the M*N quantization matrix, where: the N*M quantization matrix is a transposed matrix of an M*N quantization matrix.
  • S201 performs entropy decoding on the received code stream to obtain image data and a quantization matrix, including: entropy decoding the received code stream to obtain image data, quantization matrix, and M*N difference.
  • Quantizing the matrix; S203 obtaining, by the M*N quantization matrix, the N*M quantization matrix by transposition comprises: obtaining the N by the sum of the transposed matrix of the M*N quantization matrix and the N*M difference quantization matrix *M quantization matrix.
  • QMi a quantization matrix suitable for the current image content according to the image content.
  • S201 entropy decodes the received code stream, and the obtained quantization matrix includes QM1-QM8 and QMI 1-QM12;
  • S203 includes: QM7 and QM8 obtained by decoding QM9 and QM10 were obtained by transposition.
  • An embodiment of the present invention further provides an image encoding method.
  • Figure 3 provides a flow chart of one embodiment of the method. The method includes:
  • S303 performs transform coding on the image that is predictively encoded.
  • S305 Perform quantization coding on the transform-coded image by using a quantization matrix, where the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes an M*N quantization matrix, a P*Q quantization matrix, and the P *Q quantization matrix is obtained by scaling from the M*N quantization matrix; S307 performs entropy coding on the quantized image, and encodes the M*N quantization matrix to generate a code stream.
  • the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes an M*N quantization matrix, a P*Q quantization matrix, and the P *Q quantization matrix is obtained by scaling from the M*N quantization matrix
  • S307 performs entropy coding on the quantized image, and encodes the M*N quantization matrix to generate a code stream.
  • M is not equal to N and P is not equal to Q.
  • the *0 quantization matrix is obtained by scaling by the quantization matrix: the P*Q quantization matrix is a scaled matrix of the quantization matrix.
  • the 0 quantization matrix is obtained by scaling by the quantization matrix: the P*Q quantization matrix is predicted by the scaled matrix of the M*N quantization matrix; and the pair of the M*N quantization matrix
  • the encoding includes: calculating a P*Q difference quantization matrix according to the difference between the P*Q quantization matrix and the scaling matrix of the M*N quantization matrix, and performing the M*N quantization matrix and the P*Q difference quantization matrix coding.
  • the P*Q quantization matrix is obtained by using a scaling matrix of the M*N quantization matrix, using a lower-order interpolation, a linear interpolation, or an equally-spaced matrix weight coefficient.
  • the P*Q quantization matrix is obtained by amplifying the M*N quantization matrix
  • the quantization matrix of 8*4 and 4*8 may not be transmitted, but The 8*4, 4*8 quantization matrix is derived from the 8*8 quantization matrix.
  • the present invention proposes a method of decoding an image.
  • Figure 4 provides a flow chart of one embodiment of the method.
  • the decoding method includes:
  • S401 Entropy decoding the received code stream to obtain image data and a quantization matrix, where the quantization matrix is a matrix reflecting image quantization step size information, the quantization matrix includes an M*N quantization matrix; S403 is performed by the M*N The quantization matrix obtains a P*Q quantization matrix by scaling;
  • S405 performing inverse quantization on the entropy decoded image data by using the M*N quantization matrix and the P*Q quantization matrix;
  • the P*Q quantization matrix obtained by scaling by the M*N quantization matrix in S403 includes: the P*Q quantization matrix is a scaling matrix of an M*N quantization matrix.
  • M is not equal to N and P is not equal to Q.
  • S401 entropy decoding the received code stream to obtain image data and a quantization matrix includes: entropy decoding the received code stream to obtain image data, quantization matrix, and M*N difference.
  • Quantizing a matrix; S403 obtaining a P*Q quantization matrix by scaling by the M*N quantization matrix comprises: obtaining the P*Q by a sum of a scaling matrix of the M*N quantization matrix and the P*Q difference quantization matrix Quantization matrix.
  • the invention proposes an image encoding method.
  • Figure 5 provides a flow chart of one embodiment of the method.
  • the encoding method includes:
  • S505 performs quantization coding on the transform-coded image by using a quantization matrix, where the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes an M*N quantization matrix, a P*Q quantization matrix, and the P *Q quantization matrix is obtained by intercepting the M*N quantization matrix;
  • S507 entropy encodes the quantized image, and encodes the M*N quantization matrix to generate a code stream.
  • M is not equal to N and P is not equal to Q.
  • the P*Q quantization matrix of S505 is obtained by truncating the M*N quantization matrix to include: the P*Q quantization matrix is a truncation matrix of the M*N quantization matrix.
  • the P*Q quantization matrix of S505 is obtained by intercepting the M*N quantization matrix to include: the P*Q quantization matrix is obtained by intercepting prediction of the M*N quantization matrix;
  • the encoding of the M*N quantization matrix by S507 includes: calculating a P*Q difference quantization matrix according to a difference between the P*Q quantization matrix and the intercept matrix of the M*N quantization matrix, for M* The N quantization matrix and the P*Q difference quantization matrix are encoded.
  • the present invention provides a method of decoding an image. Please refer to Figure 6, which provides a flow chart of one embodiment of the method.
  • the decoding method includes:
  • S601 entropy decoding the received code stream to obtain image data and a quantization matrix, where the quantization matrix is a matrix reflecting image quantization step size information, the quantization matrix includes an M*N quantization matrix, and S 603 is represented by the M* The N quantization matrix is obtained by truncation to the P*Q quantization matrix;
  • S609 performs prediction compensation on the inversely transformed image data to generate a decoded image.
  • M is not equal to N and P is not equal to Q.
  • S603 is obtained by the M*N quantization matrix by truncating to the P*Q quantization matrix.
  • the P*Q quantization matrix is a truncation matrix of the M*N quantization matrix.
  • S601 entropy decoding the received code stream to obtain image data and a quantization matrix includes: entropy decoding the received code stream to obtain image data, quantization matrix, and M*N difference. Quantizing the matrix; S 603 is obtained by the M*N quantization matrix by truncating to the P* Q quantization matrix, including: obtaining the P* by the sum of the intercept matrix of the M*N quantization matrix and the P*Q difference quantization matrix Q quantization matrix.
  • Embodiments of the present invention provide an image encoding apparatus.
  • Figure 7 provides a block diagram of one embodiment of the apparatus.
  • the encoding apparatus includes: a prediction encoding module 701, configured to perform predictive encoding on an image; a transform encoding module 703, configured to transform and encode the image that is subjected to predictive encoding; and a quantization encoding module 705, configured to use a quantization matrix to transform Encoding the encoded image, the quantization matrix is a matrix reflecting image quantization step size information, the quantization matrix includes an M*N quantization matrix, an N*M quantization matrix, and the N*M quantization matrix The M*N quantization matrix is obtained by transposition; the entropy coding module 707 is configured to entropy encode the quantized image, encode the M*N quantization matrix, and generate a code stream.
  • the quantization coding module 703 is configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and an N*M quantization matrix, where the N*M quantization matrix is used by the M*N quantization matrix Obtained by transposition
  • a quantization coding module 703 configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and an N*M quantization matrix, where the N*M quantization matrix passes the M*N quantization matrix
  • the transposition prediction module is configured to calculate an N*M difference quantization matrix according to the difference between the N*M quantization matrix and the transposed matrix of the M*N quantization matrix, for the M*N
  • the quantization matrix and the N*M differential quantization matrix are encoded.
  • Embodiments of the present invention provide an image decoding apparatus.
  • Figure 8 provides a block diagram of one embodiment of the apparatus.
  • the decoding transposition includes: an entropy decoding unit 801, configured to perform entropy decoding on the received code stream to obtain image data and a quantization matrix, where the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes An M*N quantization matrix; an inverse quantization unit 803, configured to obtain an N*M quantization matrix by transposition from the M*N quantization matrix, using the M*N quantization matrix, and the N*M quantization matrix pair to pass through entropy
  • the decoded image data is inverse quantized; an inverse transform unit 805 is configured to inverse transform the inverse quantized image data; and a prediction compensation unit 807 is configured to perform prediction compensation on the inversely transformed image data to generate a decoded image.
  • the inverse quantization unit 803 is configured to assign a transposed matrix of the M*N quantization matrix to the N*M quantization matrix.
  • the entropy decoding unit 801 is configured to perform entropy decoding on the received code stream to obtain image data, a quantization matrix, and an M*N difference quantization matrix; the inverse quantization unit is used for transposition by the M*N quantization matrix The sum of the matrix and the N*M difference quantization matrix yields the N*M quantization matrix.
  • An embodiment of the present invention provides an image encoding apparatus, where the transposition includes: a prediction encoding module, configured to perform predictive encoding on an image; and a transform encoding module, configured to transform and encode the image that is subjected to predictive encoding; And an encoding module, configured to perform quantization coding on the transform-coded image by using a quantization matrix, where the quantization matrix is a matrix that reflects image quantization step size information, where the quantization matrix includes an M*N quantization matrix and a P*Q quantization matrix. The P*Q quantization matrix by the M*N The quantization matrix is obtained by scaling; an entropy coding module is configured to entropy encode the quantized image, and encode the M*N quantization matrix to generate a code stream.
  • a quantization coding module configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and a P*Q quantization matrix, where the P*Q quantization matrix is a scaled of the quantization matrix matrix.
  • a quantization coding module configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and a P*Q quantization matrix, where the P*Q quantization matrix is used by the M*N quantization matrix
  • the scaled matrix prediction is obtained;
  • the entropy coding module is configured to calculate a P*Q difference quantization matrix according to the difference between the P*Q quantization matrix and the scaling matrix of the M*N quantization matrix, and quantize the M*N
  • the matrix and the P*Q difference quantization matrix are encoded.
  • the embodiment of the present invention provides a decoding transposition of an image, where the decoding transposition includes: an entropy decoding unit, configured to perform entropy decoding on the received code stream to obtain image data and a quantization matrix, where the quantization matrix is reflected a matrix of image quantization step information, the quantization matrix includes an M*N quantization matrix; an inverse quantization unit, configured to obtain a P*Q quantization matrix by scaling from the M*N quantization matrix, using the M*N quantization matrix And the P*Q quantization matrix performs inverse quantization on the entropy-decoded image data; the inverse transform unit is configured to inverse-transform the inverse-quantized image data; and the prediction compensation unit is configured to perform inverse-transformed image data. Predictive compensation, generating a decoded image.
  • the inverse quantization unit is configured to assign a scaling matrix of the M*N quantization matrix to the P*Q quantization matrix.
  • the entropy decoding unit is configured to perform entropy decoding on the received code stream to obtain image data, a quantization matrix, and an M*N difference quantization matrix; the inverse quantization unit 803 is configured to use a scaling matrix of the M*N quantization matrix And summing the P*Q difference quantization matrix to obtain the P*Q quantization matrix.
  • An embodiment of the present invention provides an encoding transposition of an image, where the encoding transposition includes: a prediction encoding module, configured to perform predictive encoding on an image; and a transform encoding module, configured to transform and encode the image that is subjected to predictive encoding.
  • a quantization coding module configured to perform quantization coding on the transform-coded image by using a quantization matrix, where the quantization matrix reflects image quantization step information a matrix, the quantization matrix includes an M*N quantization matrix, a P*Q quantization matrix, and the P*Q quantization matrix is obtained by intercepting the M*N quantization matrix; an entropy coding module, configured to perform quantization and coding The image is entropy encoded, and the M*N quantization matrix is encoded to generate a code stream.
  • the quantization coding module is configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and a P*Q quantization matrix, where the P*Q quantization matrix is the M*N quantization The intercept matrix of the matrix.
  • the quantization coding module is configured to perform quantization coding on the transform-coded image by using a quantization matrix including an M*N quantization matrix and a P*Q quantization matrix, where the P*Q quantization matrix is quantized by the M*N
  • the interception prediction of the matrix is obtained;
  • the entropy coding module is configured to calculate a P*Q difference quantization matrix according to the difference between the P*Q quantization matrix and the intercept matrix of the M*N quantization matrix, and the M*N quantization matrix And encoding the P*Q difference quantization matrix.
  • the embodiment of the present invention provides a decoding transposition of an image, where the decoding transposition includes: an entropy decoding unit, configured to perform entropy decoding on the received code stream to obtain image data and a quantization matrix, where the quantization matrix is reflected a matrix of image quantization step information, the quantization matrix includes an M*N quantization matrix; an inverse quantization unit, configured to obtain a P*Q quantization matrix by truncation from the M*N quantization matrix, and use the M*N quantization matrix And the P*Q quantization matrix performs inverse quantization on the entropy-decoded image data; the inverse transform unit is configured to inverse-transform the inverse-quantized image data; and the prediction compensation unit is configured to perform inverse-transformed image data. Predictive compensation, generating a decoded image.
  • the inverse quantization unit is configured to assign a truncation matrix of the M*N quantization matrix to the P*Q quantization matrix.
  • the entropy decoding unit is configured to perform entropy decoding on the received code stream to obtain image data, a quantization matrix, and an M*N difference quantization matrix; and the P*Q quantization matrix is obtained by truncating the M*N quantization matrix by:
  • the inverse quantization unit is configured to obtain the P*Q quantization matrix from a sum of a truncation matrix of the M*N quantization matrix and the P*Q difference quantization matrix.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

本发明实施例提供了一种图像的编码方法,所述编码方法包括:对图像进行预测编码;对经过预测编码的所述图像进行变换编码;使用量化矩阵对变换编码后的所述图像进行量化编码,所述量化矩阵是反映图像量化步长信息的矩阵,所述量化矩阵包括M*N量化矩阵、N*M量化矩阵,所述 N*M 量化矩阵由所述M*N量化矩阵通过转置得到;对量化编码后的所述图像进行熵编码,对所述M*N量化矩阵编码,生成码流。本发明有效地节省了编码量化矩阵所需的比特数量,提高了压缩效率。

Description

一种图像编码、 解码的方法和装置 技术领域 本发明属于通信领域, 具体涉及到一种图像编码、 解码的方法和装置。 背景技术
由于视频数据量巨大,在实际应用中,视频数据通常需要压缩编码处理。 编码器通过预测、 变换、 量化和熵编码过程处理视频数据,以实现数据压缩 生成视频码流。 视频码流可用于存储或者网络传输。 解码器通过熵解码、 反量化、 反变换、 预测补偿对视频码流进行解码操作, 以重建视频数据。
H. 264编码技术通过使用量化矩阵(quant i ta t ive ma t r ix, QM)实现对信 号压缩失真的精确控制。 编码器根据当前图像内容给出一组适合当前编码 图像的 QM, 然后将该 QM编码写入码流。 解码器收到带有 QM信息的码流后, 解码出 QM信息, 并利用该 QM信息解码出图像。 在 H. 264中, 每帧图像最多可 以有 8组 QM矩阵 QMi , i=l , 2, ... 8。 8组 QM矩阵分别表示在 4 *4变换时, 帧内预 测的亮度丫、 色度 Cb、 色度 Cr和帧间预测的亮度 Y、 色度 Cb、 色度 Cr六种 QM; 以及在 8*8变换时, 帧内预测和帧间预测的亮度 Y两种 QM。 由于 QM信息数据 量较大, 需要对 QM信息进行压缩编码, 以便减少用于表示 QM信息的比特数 量。 H. 264使用以下压缩方法, 分别对 6个 4*4量化矩阵和 2两个 8 *8量化矩阵 进行编码, 具体步骤如下:
第一步, 对二维量化矩阵进行扫描操作, 生成一维数据;
第二步, 对一维数据进行 DPCM编码;
第三步, 将编码后的数据进行熵编码, 写入码流。
以上方案中釆用的变换和量化都是 N*N的正方形矩阵, 当釆用非正方形 变换和量化矩阵时, 对于 N*M的量化矩阵, 按照以上量化矩阵压缩方法, 需 要很多比特表示量化矩阵, 当应用带宽很小时, 传输量化矩阵所用比特将 严重影响编码图像的质量。
发明内容
本发明实施例提供了图像编码、 解码的方法和装置, 以减少码流的传输 带宽。
本发明实施例提供了一种图像的编码方法, 所述编码方法包括: 对图像 进行预测编码; 对经过预测编码的所述图像进行变换编码; 使用量化矩阵 对变换编码后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步 长信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 N*M量化矩阵,所述 N*M 量化矩阵由所述 M*N量化矩阵通过转置得到; 对量化编码后的所述图像进 行熵编码, 对所述 M*N量化矩阵编码, 生成码流。
本发明实施例提供了一种图像的解码方法, 所述解码方法包括: 对接 收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量化矩阵是反映 图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; 由所述 M*N 量化矩阵通过转置得到 N*M量化矩阵; 使用所述 M*N量化矩阵、 所述 N*M 量化矩阵对经过熵解码的图像数据进行反量化; 对经过反量化的图像数据 进行反变换; 对经过反变换的图像数据进行预测补偿, 生成解码图像。
本发明实施例提供了一种图像的编码装置, 所述编码装置包括: 预测 编码模块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编 码的所述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换编 码后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长信息的 矩阵, 所述量化矩阵包括 M*N量化矩阵、 N*M量化矩阵, 所述 N*M量化矩 阵由所述 M*N量化矩阵通过转置得到; 熵编码模块, 用于对量化编码后的 所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。 本发明实施例提供了一种图像的解码装置, 所述解码转置包括: 熵解 码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所 述量化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩 阵; 反量化单元, 用于由所述 M*N量化矩阵通过转置得到 N*M量化矩阵, 使 用所述 M*N量化矩阵、 所述 N*M量化矩阵对经过熵解码的图像数据进行反量 化; 反变换单元, 用于对经过反量化的图像数据进行反变换; 预测补偿单 元, 用于对经过反变换的图像数据进行预测补偿, 生成解码图像。
本发明实施例有效地节省了编码量化矩阵所需的比特数量, 提高了压 缩效率。 附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对 实施例或现有技术描述中所需要使用的附图作一简单地介绍, 显而易见, 下面描述中的附图是本发明的一些实施例, 对于本领域普通技术人员来讲, 在不付出创造性劳动的前提下, 还可以根据这些附图获得其他的附图。
图 1为本发明图像编码方法的一个实施例的流程图;
图 2为本发明图像解码方法的一个实施例的流程图;
图 3为本发明图像编码方法的又一个实施例的流程图;
图 4为本发明图像解码方法的又一个实施例的流程图;
图 5为本发明图像编码方法的再一个实施例的流程图;
图 6为本发明图像解码方法的再一个实施例的流程图;
图 7为本发明图像编码装置的一个实施例的结构图;
图 8为本发明图像解码装置的一个实施例的结构图。 具体实施方式 为使本发明实施例的目的、 技术方案和优点更加清楚, 下面将结合本 发明实施例中的附图, 对本发明实施例中的技术方案进行清楚、 完整地描 述, 显然, 所描述的实施例是本发明一部分实施例, 而不是全部的实施例。 基于本发明中的实施例, 本领域普通技术人员在没有做出创造性劳动的前 提下所获得的所有其他实施例, 都属于本发明保护的范围。
本发明实施例提供了一种图像的编码方法。 请参考附图 1 , 图 1提供了 本方法一个实施例的流程图。 所述编码方法包括:
S 1 01对图像进行预测编码;
S 1 03对经过预测编码的所述图像进行变换编码;
S 1 05使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化 矩阵是反映图像量化步长信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 N*M 量化矩阵, 所述 N*M量化矩阵由所述 M*N量化矩阵通过转置得到;
S 1 07对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。
在图像的编解码过程中, 把每一帧图像分割成较小的图像块进行编码 或者解码处理, 比如将一帧图像分割成大小为 NxN的图像块。
在本发明的一个实施例中, S 1 01具体包括: 在对 NxN的图像块编码时, NxN的图像按照不同的子块进行预测编码。 对于平坦的图像区域, 釆用较大 的图像子块划分, 以节省用于描述图像划分关系的描述信息; 对于细节丰 富的图像区域, 釆用较小的图像子块划分, 以提高图像编码时的预测准确 度; 通过自适应划分提高压缩效率。 根据预测关系的不同, 预测编码包括 帧内编码和帧间编码。 帧内编码使用来自同一帧图像的空间相邻重建像素 对当前编码块预测。 帧间预测使用当前编码块所在图像前面时刻或者后面 时刻图像的像素预测当前编码块。
在本发明的一个实施例中, S 1 03具体包括: 变换编码通过对经过预测 编码后的数据进行变换编码, 以集中数据能量和减少用于描述数据的比特 数量。 变换编码技术包括 DCT、 DST、 小波变换等。 对于不同的子块大小, 釆用不同的变换矩阵进行变换编码。 以 DCT变换为例, 对于 4x4的子块, 进 行 4x4的 DCT变换, 8x8的子块进行 8x8的 DCT变换, 32x32的子块烤肉进行 32x32的 DCT变换, 也可以进行 4x4或者 8x8的变换; 或者进行非正方形的变 换, 32x8, 16x4等。 非正方形变换优点在于可更好的适配图像内容, 提高编 码效率。
在本发明的一个实施例中, S105具体包括: 在编码一段视频信号时, 根据图像内容的不同, 选择适合该图像内容的变换矩阵, 使用量化矩阵对 变换编码后的所述图像进行量化编码, 同时将图像数据信息和变换矩阵编 码写入码流,以控制视频数据的压缩效率。所述量化矩阵包括 M*N量化矩阵、 N*M量化矩阵, 所述 N*M量化矩阵由所述 M*N量化矩阵通过转置得到。
进一步的, 在本发明的又一个实施例中, 所述 N*M量化矩阵是所述 M*N 量化矩阵的转置矩阵。
在一个同时存在 8*8、 8*4、 4*8、 4*4四种量化矩阵的发明实施方案中, 首先, 编码器根据图像内容选择适合当前图像内容的量化矩阵 QMi 中 i = 1, 2, ... , 12. QM1-QM6分别为 4*4变换矩阵帧间和帧内预测编码时 Y, Cb , Cr 分量的量化矩阵; QM7_ QM10分别为 8*4和 4*8变换矩阵帧间和帧内预测编码 时 Y分量的量化矩阵; QM11-QM12分别为 8*8变换矩阵帧间和帧内预测编码时 Y分量的量化矩阵。 其中 QM7和 QM9为转置关系, QM8和 QM10为转置关系。 接 下来, 利用量化矩阵对当前图像变换后的图像进行量化和编码, 写入码流。 同时对量化矩阵进行编码。 对量化矩阵编码时, 仅仅编码 QM1-8和 QM11-12 , 不编码 QM9和 10。 QM9和 10的信息通过解码端对 QM7和 8的信息进行转置得到。
在本发明的又一个实施例中, 所述 N*M量化矩阵由所述 M*N量化矩阵通 过转置得到包括: 根据所述 N*M量化矩阵和所述 M*N量化矩阵的转置矩阵的 差值计算得到 N*M差异量化矩阵; 所述对所述 M*N量化矩阵编码包括: 对量 化编码后的所述图像、 M*N量化矩阵和所述 M*N差异量化矩阵进行熵编码。 在本发明的一个实施例中, S107具体包括, 对每个量化矩阵可以按照 口下方法进行编码:
对二维量化矩阵进行扫描操作, 生成一维数据;
对一维数据进行 DPCM预测, 生成预测差异数据;
将预测差异数据进行熵编码, 写入码流。
在本发明的又一个实施例中, S107具体包括, 对每个量化矩阵可以按 照如下方法进行编码:
对大小为 M*N的量化矩阵 (^^按照如下方法进行编码:
对当前 (^^进行预测编码, 其预测信号来自于对 QMNxM进行转置后的信 号, 得到预测后的量化矩阵差异信号 DQMMxN
对 DQMMxN进行量化处理和扫描, 得到一维系数;
将一维系数进行熵编码, 写入码流。
本发明实施例提供了一种图像的解码方法。 请参考附图 2 , 图 2提供了 本方法一个实施例的流程图。 所述解码方法包括:
S201对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵;
S 203由所述 M*N量化矩阵通过转置得到 N*M量化矩阵;
S205使用所述 M*N量化矩阵、 所述 N*M量化矩阵对经过熵解码的图像数 据进行反量化;
S207对经过反量化的图像数据进行反变换;
S209对经过反变换的图像数据进行预测补偿, 生成解码图像。
在本发明的一个实施例中, S201对接收的码流进行熵解码, 以得到量 化矩阵的过程包括,
熵解码量化矩阵码流, 得到一维预测后的量化系数差值信号; 对一维量化系数差值信号进行 DPCM预测补偿;
进行反扫描, 得到二维的量化矩阵。 在本发明的又一个实施例中, S201对接收的码流进行熵解码, 以得到 量化矩阵的过程包括:
熵解码 QMM 々码流, 得到一维系数;
对一维系数反扫描得到二维系数矩阵;
对二维系数进行反量化得到差异信号 DQMMxN的重建值;
利用已经解码得到的 QMNxM信号的转置信号作为预测值, 用该预测值对 0(^^进行预测补偿, 得到重建的 QMMxN信号, 完成对 QMMxN的解码过程。
在本发明的一个实施例中, S203由所述 M*N量化矩阵通过转置得到 N*M 量化矩阵包括: 所述 N*M量化矩阵是 M*N量化矩阵的转置矩阵。
在本发明的又一个实施例中, S201对接收的码流进行熵解码, 以得到 图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; S203由所述 M*N量化矩阵通过转置得到 N*M量 化矩阵包括: 由所述 M*N量化矩阵的转置矩阵和所述 N*M差异量化矩阵的和 得到所述 N*M量化矩阵。
在本发明的另一个实施例中, 若编码端生成 8*8、 8*4、 4*8、 4*4四种 量化矩阵, 编码器根据图像内容选择适合当前图像内容的量化矩阵 QMi 中 i = 1, 2, ... , 12。 对量化矩阵编码时, 仅仅编码 QM1-QM8和 QM11-QM12 , S201 对接收的码流进行熵解码 ,得到的量化矩阵包括 QM1-QM8和 QMI 1-QM12 ; S203 包括: 对解码得到的 QM7和 QM8通过转置得到 QM9和 QM10。
本发明实施例还提供了一种图像的编码方法。 请参考附图 3 , 图 3提供 了本方法一个实施例的流程图。 所述方法包括:
S 301对图像进行预测编码;
S303对经过预测编码的所述图像进行变换编码;
S305使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化 矩阵是反映图像量化步长信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 P*Q 量化矩阵, 所述 P*Q量化矩阵由所述 M*N量化矩阵通过缩放得到; S307对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。
在本发明的实施例中, M不等于 N, P不等于 Q。
在本发明的一个实施例中, 8305中所述?*0量化矩阵由所述¾^^量化矩 阵通过缩放得到包括: 所述 P*Q量化矩阵是所述量化矩阵的缩放后的矩阵。
在本发明的又一个实施例中, 8305中所述?*0量化矩阵由所述¾^^量化 矩阵通过缩放得到包括: 所述 P*Q量化矩阵由所述 M*N量化矩阵的缩放后的 矩阵预测得到; 所述对所述 M*N量化矩阵编码包括: 根据所述 P*Q量化矩阵 和所述 M*N量化矩阵的缩放矩阵的差值计算得到 P*Q差异量化矩阵, 对 M*N量 化矩阵和所述 P*Q差异量化矩阵进行编码。
若 P小于 M或者 Q小于 N, 通过对所述 M*N量化矩阵的缩放矩阵釆用下釆样 插值、 线形插值或等间隔抽取矩阵重系数得到所述 P*Q量化矩阵。
若 P大于 M或者 Q大于 N, 通过放大所述 M*N量化矩阵得到所述 P*Q量化矩 阵
若在本发明的一个实施例中, 若编码端生成 8*8、 8*4、 4*8、 4*4四种 量化矩阵,可以不传输 8*4、 4* 8的量化矩阵,而是从 8* 8的量化矩阵导出 8*4、 4*8的量化矩阵。
本发明提出了一种图像的解码方法。 请参考附图 4 , 图 4提供了本方法 一个实施例的流程图。 所述解码方法包括:
S401对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; S403由所述 M*N量化矩阵通过缩放得到 P *Q量化矩阵;
S405使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数 据进行反量化;
S407对经过反量化的图像数据进行反变换;
S409对经过反变换的图像数据进行预测补偿, 生成解码图像。 在本发明的一个实施例中 , S403所述由所述 M*N量化矩阵通过缩放得到 P * Q量化矩阵包括: 所述 P * Q量化矩阵是 M*N量化矩阵的缩放矩阵。
在本发明的实施例中, M不等于 N, P不等于 Q。
在本发明的又一个实施例中, S401对接收的码流进行熵解码, 以得到 图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; S403由所述 M*N量化矩阵通过缩放得到 P*Q量 化矩阵包括: 由所述 M*N量化矩阵的缩放矩阵和所述 P*Q差异量化矩阵的和 得到所述 P*Q量化矩阵。
本发明提出了一种图像的编码方法。 请参考附图 5 , 图 5提供了本方法 一个实施例的流程图。 所述编码方法包括:
S 501对图像进行预测编码;
S503对经过预测编码的所述图像进行变换编码;
S505使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化 矩阵是反映图像量化步长信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 P*Q 量化矩阵, 所述 P*Q量化矩阵由所述 M*N量化矩阵通过截取得到;
S507对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。
在本发明的实施例中, M不等于 N, P不等于 Q。
在本发明的一个实施例中, S505所述 P*Q量化矩阵由所述 M*N量化矩阵 通过截取得到包括: 所述 P*Q量化矩阵是所述 M*N量化矩阵的截取矩阵。
在本发明的一个实施例中, S505所述 P*Q量化矩阵由所述 M*N量化矩阵 通过截取得到包括: 所述 P*Q量化矩阵通过所述 M*N量化矩阵的截取预测得 到; S507所述对所述 M*N量化矩阵编码包括: 4艮据所述 P*Q量化矩阵和所述 M*N量化矩阵的截取矩阵的差值计算得到 P*Q差异量化矩阵, 对 M*N量化矩阵 和所述 P*Q差异量化矩阵进行编码。 本发明提供了一种图像的解码方法。 请参考附图 6 , 图 6提供了本方法 一个实施例的流程图。 所述解码方法包括:
S601对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; S 603由所述 M*N量化矩阵通过截取得到 P *Q量化矩阵;
S605使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数 据进行反量化;
S607对经过反量化的图像数据进行反变换;
S609对经过反变换的图像数据进行预测补偿, 生成解码图像。
在本发明的实施例中, M不等于 N, P不等于 Q。
在本发明的一个实施例中, S603由所述 M*N量化矩阵通过截取得到 P*Q 量化矩阵包括: 所述 P*Q量化矩阵是 M*N量化矩阵的截取矩阵。
在本发明的另一个实施例中, S601对接收的码流进行熵解码, 以得到 图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; S 603由所述 M*N量化矩阵通过截取得到 P* Q量 化矩阵包括: 由所述 M*N量化矩阵的截取矩阵和所述 P*Q差异量化矩阵的和 得到所述 P*Q量化矩阵。
本发明实施例提供了一种图像的编码装置。 请参考附图 7 , 图 7提供了 本装置一个实施例的结构图。 所述编码装置包括: 预测编码模块 701 , 用于 对图像进行预测编码; 变换编码模块 703 , 用于对经过预测编码的所述图像 进行变换编码; 量化编码模块 705 , 用于使用量化矩阵对变换编码后的所述 图像进行量化编码, 所述量化矩阵是反映图像量化步长信息的矩阵, 所述 量化矩阵包括 M*N量化矩阵、 N*M量化矩阵, 所述 N*M量化矩阵由所述 M*N量 化矩阵通过转置得到; 熵编码模块 707 , 用于对量化编码后的所述图像进行 熵编码, 对所述 M*N量化矩阵编码, 生成码流。 量化编码模块 703 , 用于使用包括 M*N量化矩阵、 N*M量化矩阵的量化矩 阵对变换编码后的所述图像进行量化编码, 所述 N*M量化矩阵由所述 M*N量 化矩阵通过转置得到
量化编码模块 703 , 用于使用包括 M*N量化矩阵、 N*M量化矩阵的量化矩 阵对变换编码后的所述图像进行量化编码, 所述 N*M量化矩阵通过所述 M*N 量化矩阵的转置预测得到; 所述熵编码模块 707用于根据所述 N*M量化矩阵 和所述 M*N量化矩阵的转置矩阵的差值计算得到 N*M差异量化矩阵, 对 M*N量 化矩阵和所述 N*M差异量化矩阵进行编码。
本发明实施例提供了一种图像的解码装置。 请参考附图 8 , 图 8提供了 本装置一个实施例的结构图。 所述解码转置包括: 熵解码单元 801 , 用于对 接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量化矩阵是反 映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; 反量化单元 803 , 用于由所述 M*N量化矩阵通过转置得到 N*M量化矩阵, 使用所述 M*N量 化矩阵、 所述 N*M量化矩阵对经过熵解码的图像数据进行反量化; 反变换单 元 805 , 用于对经过反量化的图像数据进行反变换; 预测补偿单元 807 , 用 于对经过反变换的图像数据进行预测补偿, 生成解码图像。
所述反量化单元 803用于把 M*N量化矩阵的转置矩阵赋值给所述 N*M量 化矩阵。
所述熵解码单元 801用于对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; 所述反量化单元用于由所述 M*N量化矩阵的 转置矩阵和所述 N*M差异量化矩阵的和得到所述 N*M量化矩阵。
本发明实施例提供了一种图像的编码装置, 所述转置包括: 预测编码 模块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编码的 所述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换编码后 的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵,所述 P*Q量化矩阵由所述 M*N 量化矩阵通过缩放得到; 熵编码模块, 用于对量化编码后的所述图像进行 熵编码, 对所述 M*N量化矩阵编码, 生成码流。
量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵 对变换编码后的所述图像进行量化编码, 所述 P*Q量化矩阵是所述量化矩阵 的缩放后的矩阵。
量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵 对变换编码后的所述图像进行量化编码, 所述 P*Q量化矩阵由所述 M*N量化 矩阵的缩放后的矩阵预测得到; 所述熵编码模块用于根据所述 P*Q量化矩阵 和所述 M*N量化矩阵的缩放矩阵的差值计算得到 P*Q差异量化矩阵, 对 M*N量 化矩阵和所述 P*Q差异量化矩阵进行编码。
本发明实施例提供了一种图像的解码转置, 所述解码转置包括: 熵解 码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所 述量化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩 阵; 反量化单元, 用于由所述 M*N量化矩阵通过缩放得到 P*Q量化矩阵, 使 用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数据进行反量 化; 反变换单元, 用于对经过反量化的图像数据进行反变换; 预测补偿单 元, 用于对经过反变换的图像数据进行预测补偿, 生成解码图像。
所述反量化单元用于把所述 M*N量化矩阵的缩放矩阵赋值给所述 P*Q量 化矩阵。
所述熵解码单元用于对接收的码流进行熵解码, 以得到图像数据、 量化 矩阵和 M*N差异量化矩阵; 所述反量化单元 803用于由所述 M*N量化矩阵的缩 放矩阵和所述 P*Q差异量化矩阵的和得到所述 P*Q量化矩阵。
本发明实施例提供了一种图像的编码转置, 所述编码转置包括: 预测 编码模块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编 码的所述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换编 码后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长信息的 矩阵, 所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵, 所述 P*Q量化矩阵由 所述 M*N量化矩阵通过截取得到; 熵编码模块, 用于对量化编码后的所述图 像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。
所述量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化 矩阵对变换编码后的所述图像进行量化编码, 所述 P*Q量化矩阵是所述 M*N 量化矩阵的截取矩阵。
所述量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化 矩阵对变换编码后的所述图像进行量化编码, 所述 P*Q量化矩阵通过所述 M*N量化矩阵的截取预测得到; 所述熵编码模块用于根据所述 P*Q量化矩阵 和所述 M*N量化矩阵的截取矩阵的差值计算得到 P*Q差异量化矩阵, 对 M*N量 化矩阵和所述 P*Q差异量化矩阵进行编码。
本发明实施例提供了一种图像的解码转置, 所述解码转置包括: 熵解 码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所 述量化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩 阵; 反量化单元, 用于由所述 M*N量化矩阵通过截取得到 P*Q量化矩阵, 使 用所述 M*N量化矩阵、 所述 P *Q量化矩阵对经过熵解码的图像数据进行反量 化; 反变换单元, 用于 对经过反量化的图像数据进行反变换; 预测补偿单 元, 用于 对经过反变换的图像数据进行预测补偿, 生成解码图像。
所述反量化单元用于把所述 M*N量化矩阵的截取矩阵赋值给所述 P*Q 量化矩阵。
所述熵解码单元用于对接收的码流进行熵解码, 以得到图像数据、 量 化矩阵和 M*N差异量化矩阵; 由所述 M*N量化矩阵通过截取得到 P *Q量化矩阵 包括: 所述反量化单元用于由所述 M*N量化矩阵的截取矩阵和所述 P*Q差异 量化矩阵的和得到所述 P * Q量化矩阵。
最后应说明的是: 以上实施例仅用以说明本发明的技术方案, 而非对 其限制; 尽管参照前述实施例对本发明进行了详细的说明, 本领域的普通 技术人员应当理解: 其依然可以对前述各实施例所记载的技术方案进行修 改, 或者对其中部分技术特征进行等同替换; 而这些修改或者替换, 并不 使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。

Claims

权利要求
、 一种图像的编码方法, 其特征在于, 所述编码方法包括:
对图像进行预测编码;
对经过预测编码的所述图像进行变换编码;
使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化矩阵 是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵、 N*M量化矩阵, 所述 N*M量化矩阵由所述 M*N量化矩阵通过转置得到; 对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生 成码流。
、 根据权利要求 1所述的编码方法, 其特征在于, 所述 N*M量化矩阵由所述
M*N量化矩阵通过转置得到包括: 所述 N*M量化矩阵是所述 M*N量化矩 阵的转置矩阵。
、 根据权利要求 1所述的编码方法, 其特征在于, 所述 N*M量化矩阵由所述
M*N量化矩阵通过转置得到包括: 所述 N*M量化矩阵通过所述 M*N量化 矩阵的转置预测得到; 所述对所述 M*N量化矩阵编码包括: 根据所述 N*M量化矩阵和所述 M*N量化矩阵的转置矩阵的差值计算得到 N*M差异 量化矩阵, 对 M*N量化矩阵和所述 N*M差异量化矩阵进行熵编码。
、 一种图像的解码方法, 其特征在于, 所述解码方法包括:
对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量化矩 阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; 由所述 M*N量化矩阵通过转置得到 N*M量化矩阵;
使用所述 M*N量化矩阵、 所述 N*M量化矩阵对经过熵解码的图像数据进 行反量化;
对经过反量化的图像数据进行反变换;
对经过反变换的图像数据进行预测补偿, 生成解码图像。
、 根据权利要求 4所述的解码方法, 其特征在于, 由所述 M*N量化矩阵通过 转置得到 N*M量化矩阵包括: 所述 N*M量化矩阵是 M*N量化矩阵的转置 矩阵。
、 根据权利要求 4所述的解码方法, 其特征在于, 对接收的码流进行熵解 码, 以得到图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以 得到图像数据、 量化矩阵和 M*N差异量化矩阵; 所述由所述 M*N量化矩 阵通过转置得到 N*M量化矩阵包括: 由所述 M*N量化矩阵的转置矩阵和 所述 N*M差异量化矩阵的和得到所述 N*M量化矩阵。
、 一种图像的编码方法, 其特征在于, 所述方法包括:
对图像进行预测编码;
对经过预测编码的所述图像进行变换编码;
使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化矩阵 是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵, 所述 P*Q量化矩阵由所述 M*N量化矩阵通过缩放得到; 对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生 成码流。
、 根据权利要求 7所述的方法,其特征在于,所述 P*Q量化矩阵由所述 M*N量 化矩阵通过缩放得到包括: 所述 P*Q量化矩阵是所述量化矩阵的缩放 后的矩阵。
、 根据权利要求 7所述的方法,其特征在于, 所述 P*Q量化矩阵由所述 M*N量 化矩阵通过缩放得到包括: 所述 P*Q量化矩阵由所述 M*N量化矩阵的缩 放后的矩阵预测得到; 所述对所述 M*N量化矩阵编码包括: 根据所述 P*Q量化矩阵和所述 M*N量化矩阵的缩放矩阵的差值计算得到 P*Q差异 量化矩阵, 对 M*N量化矩阵和所述 P*Q差异量化矩阵进行编码。
0、 一种图像的解码方法, 其特征在于: 所述解码方法包括:
对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量化矩 阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; 由所述 M*N量化矩阵通过缩放得到 P*Q量化矩阵;
使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数据进 行反量化;
对经过反量化的图像数据进行反变换;
对经过反变换的图像数据进行预测补偿, 生成解码图像。
1 1、 根据权利要求 1 0所述的解码方法, 其特征在于, 由所述 M*N量化矩阵 通过缩放得到 P*Q量化矩阵包括: 所述 P*Q量化矩阵是 M*N量化矩阵的 缩放矩阵。
12、 根据权利要求 1 0所述的解码方法, 其特征在于, 对接收的码流进行熵 解码, 以得到图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; 由所述 M*N量化矩阵 通过缩放得到 P*Q量化矩阵包括: 由所述 M*N量化矩阵的缩放矩阵和所 述 P*Q差异量化矩阵的和得到所述 P*Q量化矩阵。
1 3、 一种图像的编码方法, 其特征在于, 所述编码方法包括:
对图像进行预测编码;
对经过预测编码的所述图像进行变换编码;
使用量化矩阵对变换编码后的所述图像进行量化编码, 所述量化矩阵 是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵, 所述 P*Q量化矩阵由所述 M*N量化矩阵通过截取得到; 对量化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生 成码流。
14、 根据权利要求 1 3所述的编码方法, 其特征在于, 所述 P*Q量化矩阵由 所述 M*N量化矩阵通过截取得到包括: 所述 P*Q量化矩阵是所述 M*N量 化矩阵的截取矩阵。
15、 根据权利要求 1 3所述的编码方法, 其特征在于, 所述 P*Q量化矩阵由 所述 M*N量化矩阵通过截取得到包括: 所述 P*Q量化矩阵通过所述 M*N 量化矩阵的截取预测得到; 所述对所述 M*N量化矩阵编码包括: 根据 所述 P*Q量化矩阵和所述 M*N量化矩阵的截取矩阵的差值计算得到 P*Q 差异量化矩阵, 对 M*N量化矩阵和所述 P*Q差异量化矩阵进行编码。 、 一种图像的解码方法, 其特征在于, 所述解码方法包括:
对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量化矩 阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化矩阵; 由所述 M*N量化矩阵通过截取得到 P*Q量化矩阵;
使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数据进 行反量化;
对经过反量化的图像数据进行反变换;
对经过反变换的图像数据进行预测补偿, 生成解码图像。
、 根据权利要求 16所述的解码方法, 其特征在于, 由所述 M*N量化矩阵 通过截取得到 P*Q量化矩阵包括: 所述 P*Q量化矩阵是 M*N量化矩阵的 截取矩阵。
、 根据权利要求 16所述的解码方法, 其特征在于, 对接收的码流进行熵 解码, 以得到图像数据和量化矩阵包括: 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量化矩阵; 由所述 M*N量化矩阵 通过截取得到 P*Q量化矩阵包括: 由所述 M*N量化矩阵的截取矩阵和所 述 P*Q差异量化矩阵的和得到所述 P*Q量化矩阵。
、 一种图像的编码装置, 其特征在于, 所述编码装置包括: 预测编码模 块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编码 的所述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换 编码后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长 信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 N*M量化矩阵,所述 N*M 量化矩阵由所述 M*N量化矩阵通过转置得到; 熵编码模块, 用于对量 化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码 Ί。
、 根据权利要求 20所述的编码装置, 其特征在于, 所述量化编码模块, 用于使用包括 Μ*Ν量化矩阵、 Ν*Μ量化矩阵的量化矩阵对变换编码后的 所述图像进行量化编码, 所述 Ν*Μ量化矩阵由所述 Μ*Ν量化矩阵通过转 置得到
、 根据权利要求 20所述的编码装置, 其特征在于, 所述量化编码模块, 用于使用包括 Μ*Ν量化矩阵、 Ν*Μ量化矩阵的量化矩阵对变换编码后的 所述图像进行量化编码, 所述 Ν*Μ量化矩阵通过所述 Μ*Ν量化矩阵的转 置预测得到; 所述熵编码模块用于 4艮据所述 Ν*Μ量化矩阵和所述 Μ*Ν量 化矩阵的转置矩阵的差值计算得到 Ν*Μ差异量化矩阵,对 Μ*Ν量化矩阵 和所述 Ν*Μ差异量化矩阵进行编码。
、 一种图像的解码装置,其特征在于,所述解码转置包括: 熵解码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 Μ*Ν量化 矩阵; 反量化单元, 用于由所述 Μ*Ν量化矩阵通过转置得到 Ν*Μ量化矩 阵, 使用所述 Μ*Ν量化矩阵、 所述 Ν*Μ量化矩阵对经过熵解码的图像数 据进行反量化; 反变换单元, 用于对经过反量化的图像数据进行反变 换; 预测补偿单元, 用于对经过反变换的图像数据进行预测补偿, 生 成解码图像。
、 根据权利要求 22所述的解码装置, 其特征在于, 所述反量化单元用于 4巴 Μ*Ν量化矩阵的转置矩阵!!武值给所述 Ν*Μ量化矩阵。
、 根据权利要求 22所述的解码装置, 其特征在于, 所述熵解码单元用于 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 Μ*Ν差异量 化矩阵; 所述反量化单元用于由所述 Μ*Ν量化矩阵的转置矩阵和所述 Ν*Μ差异量化矩阵的和得到所述 Ν*Μ量化矩阵。
、 一种图像的编码装置, 其特征在于, 所述转置包括: 预测编码模块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编码的所 述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换编码 后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长信息 的矩阵, 所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵, 所述 P*Q量 化矩阵由所述 M*N量化矩阵通过缩放得到; 熵编码模块, 用于对量化 编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码流。 、 根据权利要求 25所述的转置,其特征在于, 所述所述量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵对变换编码后的 所述图像进行量化编码, 所述 P * Q量化矩阵是所述量化矩阵的缩放后 的矩阵。
、 根据权利要求 25所述的转置,其特征在于, 所述量化编码模块, 用于 使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵对变换编码后的所述 图像进行量化编码, 所述 P*Q量化矩阵由所述 M*N量化矩阵的缩放后的 矩阵预测得到; 所述熵编码模块用于根据所述 P*Q量化矩阵和所述 M*N 量化矩阵的缩放矩阵的差值计算得到 P*Q差异量化矩阵,对 M*N量化矩 阵和所述 P*Q差异量化矩阵进行编码。
、 一种图像的解码转置,其特征在于: 所述解码转置包括: 熵解码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化 矩阵; 反量化单元, 用于由所述 M*N量化矩阵通过缩放得到 P*Q量化矩 阵, 使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数 据进行反量化; 反变换单元, 用于对经过反量化的图像数据进行反变 换; 预测补偿单元, 用于 对经过反变换的图像数据进行预测补偿, 生成解码图像。
、 根据权利要求 28所述的解码转置, 其特征在于, 所述反量化单元用于 把所述 M*N量化矩阵的缩放矩阵赋值给所述 P*Q量化矩阵。 、 根据权利要求 28所述的解码转置, 其特征在于, 所述熵解码单元用于 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量 化矩阵; 所述反量化单元用于由所述 M*N量化矩阵的缩放矩阵和所述
P*Q差异量化矩阵的和得到所述 P*Q量化矩阵。
、 一种图像的编码转置, 其特征在于, 所述编码转置包括: 预测编码模 块, 用于对图像进行预测编码; 变换编码模块, 用于对经过预测编码 的所述图像进行变换编码; 量化编码模块, 用于使用量化矩阵对变换 编码后的所述图像进行量化编码, 所述量化矩阵是反映图像量化步长 信息的矩阵,所述量化矩阵包括 M*N量化矩阵、 P*Q量化矩阵,所述 P*Q 量化矩阵由所述 M*N量化矩阵通过截取得到; 熵编码模块, 用于对量 化编码后的所述图像进行熵编码, 对所述 M*N量化矩阵编码, 生成码 流。
、 根据权利要求 31所述的编码转置, 其特征在于, 所述量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵对变换编码后的 所述图像进行量化编码, 所述 P*Q量化矩阵是所述 M*N量化矩阵的截取 矩阵。
、 根据权利要求 31所述的编码转置, 其特征在于, 所述量化编码模块, 用于使用包括 M*N量化矩阵、 P*Q量化矩阵的量化矩阵对变换编码后的 所述图像进行量化编码, 所述 P*Q量化矩阵通过所述 M*N量化矩阵的截 取预测得到; 所述熵编码模块用于根据所述 P*Q量化矩阵和所述 M*N量 化矩阵的截取矩阵的差值计算得到 P*Q差异量化矩阵,对 M*N量化矩阵 和所述 P*Q差异量化矩阵进行编码。
、 一种图像的解码转置,其特征在于,所述解码转置包括: 熵解码单元, 用于对接收的码流进行熵解码, 以得到图像数据和量化矩阵, 所述量 化矩阵是反映图像量化步长信息的矩阵, 所述量化矩阵包括 M*N量化 矩阵; 反量化单元, 用于由所述 M*N量化矩阵通过截取得到 P*Q量化矩 阵, 使用所述 M*N量化矩阵、 所述 P*Q量化矩阵对经过熵解码的图像数 据进行反量化; 反变换单元, 用于 对经过反量化的图像数据进行反 变换; 预测补偿单元, 用于 对经过反变换的图像数据进行预测补偿, 生成解码图像。
35、 根据权利要求 34所述的解码转置, 其特征在于, 所述反量化单元用于 把所述 M*N量化矩阵的截取矩阵赋值给所述 P*Q量化矩阵。
36、 根据权利要求 34所述的解码转置, 其特征在于, 所述熵解码单元用于 对接收的码流进行熵解码, 以得到图像数据、 量化矩阵和 M*N差异量 化矩阵; 由所述 M*N量化矩阵通过截取得到 P*Q量化矩阵包括: 所述反 量化单元用于由所述 M*N量化矩阵的截取矩阵和所述 P*Q差异量化矩 阵的和得到所述 P * Q量化矩阵。
PCT/CN2012/084062 2011-11-04 2012-11-05 一种图像编码、解码的方法和装置 WO2013064109A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP12846329.6A EP2750386A4 (en) 2011-11-04 2012-11-05 METHOD AND DEVICE FOR ENCODING AND DECODING IMAGE
US14/268,115 US9667958B2 (en) 2011-11-04 2014-05-02 Image coding and decoding methods and apparatuses
US15/492,868 US10091531B2 (en) 2011-11-04 2017-04-20 Image coding and decoding methods and apparatuses

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110345953.8 2011-11-04
CN201110345953.8A CN103096052B (zh) 2011-11-04 2011-11-04 一种图像编码、解码的方法和装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/268,115 Continuation US9667958B2 (en) 2011-11-04 2014-05-02 Image coding and decoding methods and apparatuses

Publications (1)

Publication Number Publication Date
WO2013064109A1 true WO2013064109A1 (zh) 2013-05-10

Family

ID=48191369

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2012/084062 WO2013064109A1 (zh) 2011-11-04 2012-11-05 一种图像编码、解码的方法和装置

Country Status (4)

Country Link
US (2) US9667958B2 (zh)
EP (1) EP2750386A4 (zh)
CN (1) CN103096052B (zh)
WO (1) WO2013064109A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017075810A1 (zh) * 2015-11-06 2017-05-11 华为技术有限公司 反量化变换系数的方法、装置及解码设备
WO2022027442A1 (zh) * 2020-08-06 2022-02-10 华为技术有限公司 图像处理网络的输入预处理方法和输出后处理方法及装置

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10200719B2 (en) * 2015-11-25 2019-02-05 Qualcomm Incorporated Modification of transform coefficients for non-square transform units in video coding
JP2020098983A (ja) * 2018-12-17 2020-06-25 キヤノン株式会社 画像符号化装置、画像符号化方法、画像復号装置、画像復号方法
CN114079772B (zh) * 2020-08-21 2023-04-07 腾讯科技(深圳)有限公司 视频解码方法、装置、计算机可读介质及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101039421A (zh) * 2006-03-16 2007-09-19 华为技术有限公司 在编解码中的实现量化的方法和装置
CN101194510A (zh) * 2005-02-08 2008-06-04 德州仪器公司 H.264量化
CN101438591A (zh) * 2006-05-05 2009-05-20 微软公司 灵活量化
CN101946513A (zh) * 2007-12-11 2011-01-12 三星电子株式会社 用于量化的方法和设备以及用于反量化的方法和设备

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5682152A (en) * 1996-03-19 1997-10-28 Johnson-Grace Company Data compression using adaptive bit allocation and hybrid lossless entropy encoding
US7376280B2 (en) * 2002-07-14 2008-05-20 Apple Inc Video encoding and decoding
EP1995971B1 (en) * 2006-03-16 2016-11-16 Huawei Technologies Co., Ltd. A method and device for realizing quantization in coding-decoding
US8315204B2 (en) * 2009-07-06 2012-11-20 Intel Corporation Beamforming using base and differential codebooks
AU2010312983B2 (en) * 2009-10-30 2015-10-29 Sun Patent Trust Image decoding method, image coding method, image decoding apparatus, image coding apparatus, program, and integrated circuit
US9167252B2 (en) * 2010-12-01 2015-10-20 Texas Instruments Incorporated Quantization matrix compression in video coding
US9807426B2 (en) * 2011-07-01 2017-10-31 Qualcomm Incorporated Applying non-square transforms to video data
JP2013038768A (ja) * 2011-07-13 2013-02-21 Canon Inc 画像符号化装置、画像符号化方法及びプログラム、画像復号装置、画像復号方法及びプログラム

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101194510A (zh) * 2005-02-08 2008-06-04 德州仪器公司 H.264量化
CN101039421A (zh) * 2006-03-16 2007-09-19 华为技术有限公司 在编解码中的实现量化的方法和装置
CN101438591A (zh) * 2006-05-05 2009-05-20 微软公司 灵活量化
CN101946513A (zh) * 2007-12-11 2011-01-12 三星电子株式会社 用于量化的方法和设备以及用于反量化的方法和设备

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017075810A1 (zh) * 2015-11-06 2017-05-11 华为技术有限公司 反量化变换系数的方法、装置及解码设备
US10630983B2 (en) 2015-11-06 2020-04-21 Huawei Technologies Co., Ltd. Transform coefficient dequantization method and apparatus, and decoding device
WO2022027442A1 (zh) * 2020-08-06 2022-02-10 华为技术有限公司 图像处理网络的输入预处理方法和输出后处理方法及装置

Also Published As

Publication number Publication date
EP2750386A4 (en) 2014-08-06
CN103096052B (zh) 2015-11-25
US20140241423A1 (en) 2014-08-28
CN103096052A (zh) 2013-05-08
US9667958B2 (en) 2017-05-30
EP2750386A1 (en) 2014-07-02
US20170223381A1 (en) 2017-08-03
US10091531B2 (en) 2018-10-02

Similar Documents

Publication Publication Date Title
JP7063965B2 (ja) ビデオ符号化/復号化システムのための多次元量子化技術
KR101477302B1 (ko) 디지털 미디어 코딩/디코딩 방법, 디지털 미디어 처리 장치및 프로그램 저장 캐리어
JP2021511752A (ja) 縮小された変換を用いて、ビデオ信号を処理する方法及び装置
US11909963B2 (en) Image encoding method, decoding method, encoder, and decoder
US9942568B2 (en) Hybrid transform scheme for video coding
EP2774360A1 (en) Differential pulse code modulation intra prediction for high efficiency video coding
US10091531B2 (en) Image coding and decoding methods and apparatuses
JP2022526400A (ja) ビデオコーディングのための方法および装置
CN113491129B (zh) 图像/视频压缩中的自适应滤波器帧内预测模式
AU2020360013B2 (en) Transform-based image coding method, and apparatus therefor
US10277905B2 (en) Transform selection for non-baseband signal coding
WO2022139934A1 (en) Method and apparatus for video coding
EP3871414A1 (en) Method and apparatus for video coding
JP2022514505A (ja) ビットレート削減のためのビデオストリーム適応フィルタリング
US20220345690A1 (en) Harmonized Design Among Multiple Reference Line Intra Prediction, Transform Partitioning and Transform Kernels
WO2022220906A1 (en) Improved entropy coding of sign map for transform coefficients
CA3159801A1 (en) Image coding method based on transform, and device therefor
US20230100043A1 (en) Adaptive Transforms for Compound Inter-Intra Prediction Modes
AU2020360102B2 (en) Transform-based image coding method, and device therefor
JP7090490B2 (ja) 画像符号化装置、画像復号装置、画像符号化方法、画像復号方法
WO2024081011A1 (en) Filter coefficient derivation simplification for cross-component prediction
WO2023219722A1 (en) Systems and methods for transform coefficient sign prediction and coding
WO2023196093A1 (en) Systems and methods for frequency-dependent coefficient sign coding
WO2023244337A1 (en) Systems and methods for joint signaling of transform coefficient signs
WO2023235002A1 (en) Systems and methods for partition-based predictions

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12846329

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE