• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The invention relates to a method for decoding a digital image, from a bit stream comprising coded data representative of said image, comprising the following steps, implemented for a block of said image, said transformed current block: - Decoding the coefficients of the current block transformed from coded data read in the bitstream; Transforming the current block into a decoded block, said step implementing a first substep intended to produce an intermediate block, applying to the column vectors respectively lines of the current block, the second intended to produce a block of pixels; applying to the row vectors respectively columns of the intermediate block, resulting from the first substep; - Reconstruction of the image from the decoded block; Characterized in that: at least one of said first and second transformation substeps comprises, for a said line vector respectively column, said input vector: - the formation of at least a first sub-vector of size K < N respectively N from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed sub-vector.
    公开号:FR3050598A1
    申请号:FR1653704
    申请日:2016-04-26
    公开日:2017-10-27
    发明作者:Pierrick Philippe；Victorien Lorcy；Pierre Castel
    申请人:Orange SA；B Com SAS；
    IPC主号:
    专利说明:

    Method for decoding a digital image, coding method, devices, and associated computer programs 1. Field of the invention
    The field of the invention is that of signal compression, in particular a digital image or a sequence of digital images, divided into blocks of pixels. The invention relates more particularly to the transformation of a block of pixels, whether or not a prediction. It finds particular application in a context of competition transformations.
    The encoding / decoding of digital images applies in particular to images originating from at least one video sequence comprising: images coming from the same camera and succeeding one another temporally (coding / decoding of 2D type), images from different cameras oriented in different views (3D type coding / decoding), - corresponding texture and depth components (3D type coding / decoding), - etc.
    The present invention applies similarly to the coding / decoding of 2D or 3D type images. The invention may notably but not exclusively apply to the video coding implemented in the current video codecs AVC (for "Advanced Video Coding" in English) and HEVC (for "High Efficiency Video Coding"). and their extensions (MVC, 3D-AVC, MV-HEVC, 3D-HEVC, post-HEVC, etc.), and corresponding decoding. 2. Presentation of the prior art
    We consider a conventional compression scheme of a digital image, according to which the image is divided into blocks of pixels. A current block to be coded, which constitutes an initial coding unit, is generally divided into a variable number of sub-blocks according to a predetermined cutting mode. In relation to FIG. 1, we consider a sequence of digital images Ii, h, Ij, with a non-zero integer J. An image Ij is divided into initial coding units or CTUs (for "Coding Tree Unit" in English) according to the terminology of the HEVC standard, as specified in the document "ISO / IEC 23008-2: 2013 - High efficiency coding and High standard video coding ", International Organization for Standardization, published November 2013. Standard encoders typically provide regular partitioning, based on square or rectangular blocks, called Cü (for" Coding Units, in English) of fixed size, partitioning is always done from the initial, unpartitioned coding unit, and the final partitioning is calculated and reported from this neutral base.
    Each CU undergoes an encoding or decoding operation consisting of a sequence of operations, including in a non-exhaustive manner a prediction, a residue calculation, a transformation, a quantization and an entropy coding. This series of operations is known from the prior art and presented in connection with FIG. 2.
    During a step E0, the current block C is selected as the first block CTU to be processed. For example, this is the first block (in lexicographic order). This block is for example 64x64 size according to the HEVC standard.
    It is assumed that there are D partitions in possible CUs numbered from 1 to D, with D nonzero integer and that the partitioning used on block c corresponds to partitioning number d. For example, there may be 4 possible square partition sizes, in 4x4, 8x8, 16x16, and 32x32 sub-blocks in a regular quad-tree mode as specified in the HEVC standard. Partitioning in rectangular sub-blocks is also possible.
    In the following, we denote by current block a sub-block P from the partitioning CTU block c. The steps that will be described are repeated for the other sub-blocks. During a step E1, a prediction Pr of the block CU P is determined. It is a prediction block constructed by known means, typically by motion compensation (block resulting from a previously decoded reference image ), or by intra prediction (block built from the decoded pixels belonging to the ID image). The prediction information related to Pr is encoded in the TB bit stream or compressed FC file. We suppose here that there are P possible prediction modes mi, m2, ..., mp, with P nonzero integer. For example, the prediction mode chosen for the current block x is the mp mode. Some prediction modes are associated with an Intra type prediction, others with an INTER type prediction.
    During a step E2, an original residue R is formed by subtraction R = P-Pr from the prediction Pr of the current block P to the current block P.
    In E3, we identify a transform T to be applied to the residue R or to a sub-block resulting from a subdivision of R. The transformation step plays a crucial role in such a video coding scheme: indeed, it is it focuses the information before the quantization operation. As a result, a set of pixels before encoding is shown on a small number of non-zero frequency coefficients representing the same information. Thus, instead of transmitting a large number of coefficients, only a small number will be needed to faithfully reconstruct a block of pixels.
    This transformation step is complex to implement both coder side and decoder side, which must implement the inverse transformation of that applied by the encoder.
    During a step E4, the residue R is transformed into a transformed residue block, called RT, by the identified transform. Alternatively, in the absence of prediction, a transformed block RT is obtained from block c. This is for example a block-type transform or a wavelet transform, all known to those skilled in the art and in particular implemented in the JPEG / MPEG standards for the DCT / DST and JPEG2000 for the transform. in wavelets.
    In image and video coding, block (4x4, 8x8, etc.), orthogonal or quasi-orthogonal transforms are generally used. The most used transforms are based on cosine or sinus bases. They are generally referred to as DTT (for "Discrete Trigonometry Transforms"). The DCT is thus present in most standards for image and video. Recently, the HEVC standard has also introduced the DST (for "Discrete Sine Transform") for the coding of particular residues in the case of 4x4 blocks.
    In fact, approximations of these transforms are used, the computations being carried out on integers. In general, the bases of transforms are approximated to the nearest integer, after multiplication by a factor which conditions the precision given to the approximation (this factor is often in power of 2 generally of 8 or 10 bits). By way of example, in relation to FIGS. 2A and 2B, the transforms used by the HEVC standard on blocks of 4x4 size are presented: These are the DCT and DST transforms. The values presented in this table are to be divided by 128 to find the quasi-orthonormal transformations.
    In E5, in a manner known in the state of the art, these coefficients are traversed in a predetermined order so as to constitute a one-dimensional vector RQ [j], where the index j varies from 0 to
    Nb-1, with Nb integer equal to the number of pixels of the block x. The index j is called the frequency of the coefficient RQ [j]. Classically, these coefficients are scanned by globally increasing or decreasing order of frequency values, for example according to a predetermined path, for example diagonal or horizontal.
    In E6, the transformed block RT is quantized by conventional quantization means, for example scalar or vector, into a quantized block RQ comprising as many coefficients Nb as the block RT.
    During a step E7, the information relating to the coefficients of the block RQ is coded by entropy coding, for example according to a Huffman coding or arithmetic coding technique. This information includes at least the amplitude of the coefficients and their sign. By amplitude is meant here the absolute value of the coefficient. Conventionally, one can encode for each coefficient information representative of the fact that the coefficient is non-zero. Then, for each nonzero coefficient, one or more information relating to the amplitude is encoded. CA amplitudes are obtained. The signs of the non-harm coefficients are also coded. In general, they are simply encoded by a bit 0 or 1, each value corresponding to a given polarity. Such coding obtains effective performances because, because of the transformation, the values of the amplitudes to be coded are for the most part zero.
    Concerning the applied transform, in the case of the HEVC standard, the decoder is indicated by a bit, called "transform skip_flag", the inverse transform to be applied among the two alternatives DST or absence of transform. This case occurs in the case of 4x4 blocks.
    In E8, the coded data relating to the current block x are inserted in the bit stream TB.
    The other sub-blocks which constitute the R block of the image II are treated in the same way, then the other CTU blocks of the image II, as well as the blocks of the following images of the sequence. 2. Presentation of the Prior Art The step of transforming a current block into a transformed block generates complex calculations, which the skilled person has attempted to simplify and / or accelerate.
    This complexity increases with the size of the blocks processed. In HEVC, transforms can reach a size of 32x32.
    Of particular note is Markus Püschel's Algebraic Signal Processing Theory: Cooley-Tukey Type Algorithm for DCTs and DSTs, published in the journal IEEE Transactions on Signal Processing, in April 2008, a method for rapid implementation of the DCT transformation families or DST. This method relates in particular to DCT and DST trigonometric transforms of types I to VIII and it is shown that fast algorithms for very specific sizes (typically in power of 2 or close to a power of two) can be implemented. 3. Disadvantages of prior art
    A disadvantage of these types of DCT or DST transformations is that not all of them are applicable to all block sizes, especially those specified in the HEVC standard. For example, the DST as specified in HEVC is limited to 4x4 block sizes because it induces significant complexity for larger sizes.
    Another disadvantage is that some transforms do not have fast algorithms. 4. Objectives of the invention The invention improves the situation. The invention particularly aims to overcome these disadvantages of the prior art.
    More precisely, an objective of the invention is to propose a solution that makes it possible to reduce the complexity of the calculations implemented when applying a transform to a block of pixels, to the encoder as well as to the decoder.
    Another objective of the invention is to propose a gain in complexity that has no impact on the efficiency in compression. 5. Presentation of the invention
    These objectives, as well as others which will appear later, are achieved by means of a method of decoding a digital image, from a bit stream comprising coded data representative of said image, said image being divided into a plurality of blocks processed in a defined order, said method comprising the following steps, implemented for a block, said transformed current block: decoding of the current block coefficients from coded data read in the bitstream; Transforming the current block into a decoded block, said step implementing a first substep intended to produce an intermediate block, applying to the column vectors respectively lines of the current block, the second intended to produce a block of pixels; applying to the row vectors respectively columns of the intermediate block, resulting from the first substep; and - reconstructing the image from the transformed decoded block.
    Such a method is particular in that at least one of said first and second transformation substeps comprises, for a line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively N from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a sub-transform of size KxK; and obtaining the input vector transformed by insertion of the at least one first transformed subvector.
    We consider a separable transformation implemented from two orthogonal or quasi-orthogonal transforms. The invention is based on an entirely new and inventive approach, according to which at least one of the orthogonal transformation sub-steps is replaced by at least a partial transformation of the row or column vectors of the block to be transformed. This partial transformation applies to a sub-vector of the row or column vector to be transformed and constitutes the only transformation applied to its elements, since the coefficients resulting from this transformation are directly placed back into an output vector intended for the sub-stage. -transformation following. There is therefore no recombination of the values obtained with other data resulting from a transformation of other elements of the input vector. The fact of resorting to at least one partial transformation of size KxK with K <N or M nonzero, results in fewer computation operations than a single transformation of the entire vector, which makes it possible to reduce the complexity of decoding. .
    According to an advantageous characteristic of the invention, said sub processing step further comprises the formation of at least a second sub-vector of size less than or equal to NK or MK respectively from adjacent elements of the input vector, not included in the first sub-vector, such that the sum of the sizes of the sub-vectors formed is equal to the size of the input vector, the application of a second sub-transformed partial to said second sub-vector and in that the composition transformed vector comprises inserting the elements of the first and at least one second transformed sub-vector at the initial positions of the first and at least one second sub-vectors formed in the input vector.
    Thus, all elements of the input vectors, row or column, are each processed by one and only one sub-transform. As proved later in the description, the decomposition into partial sub-transformations proposed by the invention leads to a number of operations less than or equal to that necessary to apply a single transform of size equal to that of the input vectors.
    According to another aspect of the invention, one of the at least two partial sub-transforms is an identity transform.
    As a result, at least a subset of the pixels / coefficients of the vectors (rows or columns) are not transformed during the transformation sub-step, which has the effect of reducing the complexity of the overall transformation.
    According to yet another aspect of the invention, at least one of the subvectors of the input vector is of odd size.
    One advantage is that it can be applied to a partial sub-transform of odd size, for which we have a fast algorithm. This is the case, for example, of type DTT type DTT transforms.
    According to yet another aspect of the invention, the coding method comprises a step of selecting according to a rate-distortion criterion of a pair of sub-transforms out of a plurality of predetermined pairs. Said step is particular in that at least one sub-transform of a said pair is decomposed into at least the first and the second sub-transforms.
    In a context of competition of transforms, the encoder selects, for the current block, the transforms which realize the best compromise rate-distortion. When an association of partial subprocesses according to the invention is chosen for at least one of the first or second substeps, the resulting gain in coding complexity is added to the gain in compression.
    The method which has just been described in its different embodiments is advantageously implemented by a device for decoding a digital image, from a bit stream comprising encoded data representative of said image, said image being divided in a plurality of blocks processed in a defined order, said device comprising the following units, able to be implemented for a block, called the transformed current block: decoder of the coefficients of the current block from coded data read in the bit stream ; Transformer of the current block transformed into a transformed decoded block, said transformer being able to implement a first transforming sub-unit capable of producing an intermediate block, applying to the column vectors respectively lines of the current block, a second sub-unit transformation unit capable of producing a block of pixels applying to the row vectors respectively columns of the intermediate block, derived from the first sub-unit; and - reconstructing the image from the transformed decoded block.
    Such a device is particular in that at least one of said first and second transformation subunits comprises, for a said line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N or N respectively from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed sub-vector.
    Correlatively, the invention also relates to a coding method of a digital image, said image being divided into a plurality of blocks of pixels processed in a defined order, said method comprising the following steps, implemented for a current block of dimensions. predetermined: - Transformation of the current block into a transformed block, said current block comprising M row vectors and N column vectors, with M and N non-harmonic integers, said step comprising a first substep of transformation of the M row vectors, respectively columns, intended to provide an intermediate block formed from the row and column vectors transformed respectively and a second substep of transformation of the M column vectors, respectively rows of the intermediate block; - Encoding the transformed block intended to produce coded data representative of the transformed block; - Insertion of coded data in a bitstream representative of the coded picture.
    According to the invention, such a method is particular in that at least one of said first and second transformation substeps comprises, for a line vector respectively column, said input vector: - the formation of at least one first sub-vector of size K <N or N respectively from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed sub-vector.
    The coding method according to the invention performs the inverse transformation step of that implemented by the decoding method which has just been described with its different embodiments.
    The coding method according to the invention is advantageously implemented by a device for decoding a digital image, said image being divided into a plurality of blocks of pixels processed in a defined order, said device comprising the following units, suitable for be implemented for a current block (x), of predetermined dimensions: - Transformer of the current block into a transformed block, said current block comprising M row vectors and N column vectors, with M and N whole no harm, said transformer comprising a first subunit of transformation of the M row vectors, respectively columns, capable of providing an intermediate block formed from the respectively transformed column line vectors and a second transformation subunit of the M column vectors, respectively rows of the intermediate block; - Converted block encoder for producing coded data representative of the transformed block; - Manufacturer of a binary train representative of the coded image adapted to insert said coded data in the bit stream;
    Such a device is particular in that at least one of said first and second transformation subunits comprises, for a line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively N from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed subvector. The invention also relates to a signal carrying a bit stream comprising coded data representative of a digital image, said digital image being divided into blocks of processed pixels in a defined order, a transformed block being obtained by transforming the pixels of a block current, said transformation comprising a first substep intended to produce an intermediate block, applying to the row vectors respectively columns of the current block, the second intended to produce a block of pixels applying to the column vectors respectively lines of the intermediate block, from the first sub-step.
    Such a signal is particular in that, at least one of said first and second transformation substeps comprises the formation of a sub-vector of size K <N or N respectively from adjacent elements of a said line vector respectively column, the application of a partial sub-transform of size KxK to said sub-vector and the composition of a line vector respectively column transformed by insertion of the at least one first transformed sub-vector, said signal comprises a representative identifier of said partial sub-transform. The invention further relates to a user terminal characterized in that it comprises the coding device of a digital image and the decoding device of a digital image which have just been described. The invention also relates to a computer program comprising instructions for implementing the steps of a method of decoding a digital image as described above, when this program is executed by a processor. The invention also relates to a computer program comprising instructions for implementing the steps of a method of coding a digital image as described above, when this program is executed by a processor.
    These programs can use any programming language. They can be downloaded from a communication network and / or recorded on a computer-readable medium. The invention finally relates to recording media, readable by a processor, integrated or not integrated with the encoding device of a digital image and the decoding device of a digital image according to the invention, possibly removable, respectively memorizing a computer program implementing an encoding method and a computer program implementing a decoding method, as described above. 6. List of Figures Other advantages and features of the invention will appear more clearly on reading the following description of a particular embodiment of the invention, given as a simple illustrative and non-limiting example, and attached drawings, among which: Figure 1 (already described) schematically shows a sequence of digital images cut into blocks of pixels; FIGS. 2A and 2B (already described) show the DCT and DST transformations of size 4 implemented by the encoder of the HEVC standard; FIG. 3 (already described) shows schematically the steps of a coding method of a digital image according to the prior art; FIG. 4 presents in more detail the step of transforming a current block of a digital image according to the prior art; FIGS. 5A to 5C show examples of partial sub-transformations according to the invention; FIG. 6 schematically shows an example of a substep of transformation of the row or column vectors of a block according to one embodiment of the invention; Figure 7 schematically shows the steps of a method of decoding a digital image; FIG. 8 schematically shows the hardware structure of a coding device of a digital image according to the invention; and FIG. 9 schematically shows the hardware structure of a device for decoding a digital image according to the invention. 7. Description of a particular embodiment of the invention
    In the remainder of the description, one places oneself within the framework of a coding scheme of a digital image, for example as previously described in relation with FIG. 3.
    Note however that the prediction step of a current block El described above is optional and that the next step E4 of transformation can therefore be applied directly to the pixels of the current block.
    Consider a rectangular current block of size NxM (N columns, M lines), with N and M integers not harmful and at least one transform Ti to be applied to this block. Transform T 1 belongs to a set of L transforms {Το, Τι,.,. Τι .., TL-i>, with L nonzero integer.
    In relation to FIG. 4, an exemplary embodiment of a transformation of a block x of pixels using a transform Ti is described. It is considered, in a manner known to those skilled in the art, that the application of this transform Ti may be decomposed into two successive transformation sub-steps by transforms Bi and Al, which will now be presented.
    In E40 we extract the M line vectors Ln [m] of size N from the current block x, with m integer between 0 and M-1. In E41, the M vectors Ln [m] are transposed to be presented in the form of N-size column vectors. In E42, they are transformed by a first NxN-sized transform Bi. An intermediate block of size MxN is obtained. In E43, N column vectors of size M are extracted and then transposed into E44. Transposed vectors Col [n] with n between 0 and N-1 are transformed into E45 by a second transform Ai of size MxM.
    Thus Ai and Bi can operate in the following way:
    (1)
    Al and B are transformations that can be expressed as square matrices (respectively of size MxM and NxN) and which are respectively adapted to transform each constituent vector of the block of pixels. Xi is the block of coefficients resulting from the transformation of the block x by the transform Ti.
    It is also possible to express Xi in other equivalent forms, for example by omitting the transposition of the block x. In this case the calculation is made:
    (2)
    In this case, the N column vectors are transformed firstly by a transformation of size M, followed by a transformation acting on the result lines by M transformations of size N.
    In a known manner, transformations are typically applied to row / column vectors of sizes equal to powers of 2. For example, the HEVC standard transforms row / column vectors of sizes 4.8, 16 or 32.
    It is assumed here that the first and second transforms Ai, Bi are orthogonal or quasi-orthogonal transforms. Such transformations have the property that a transformation matrix multiplied by its transpose has only constant non-zero diagonal terms. In the quasi-orthogonal case the terms outside the diagonal have negligible values with respect to the amplitude of the diagonal terms. As a result, the inverse of an orthogonal matrix is approximated by its transpose by a factor, so that at decoding, the transpose of the matrix used in the encoding can be applied.
    It will be noted that the Al and B1 transforms can also be applied through a rapid implementation by an algorithm that is in the form of a butterfly diagram (ie linear combinations of terms taken two to three). two). This applies to all trigonometric transformations, as described in the article by Markus Püschel already cited.
    One known way to appreciate the computational complexity of such a transformation step is to count the number of addition and multiplication operations required. Table 1 below gives some examples of computational complexity estimates made using this method.
    We distinguish the "general case" where a transformation is applied in matrix form and special cases (DCT I, DCT II) which rely on fast implementations of transformations according to the article by Markus Püschel cited above. The number of reported operations is relative to the processing of a pixel vector.
    Table 1
    We consider a transformation matrix comprising N rows and N columns. In the general case, we have N multiplications and N-1 additions for each row of the matrix. As a result, we have N * N and N * (N-1) multiplications and additions. We then have a number of operations per transformed value of 2N-1.
    It is noted that the complexity increases significantly with the size of the block to be treated. We also note that there are so-called fast transforms, which are more economical in computing resources.
    In the case of quasi-orthogonal transforms, the inverse transformation returning to apply the transformation with transposed coefficients, the computational complexity for the inverse is identical to the direct transformation.
    In relation to FIG. 5, the transformation step E4 according to one embodiment of the invention is now detailed. The invention proposes modifying at least one of the sub-transformation steps E42 of the columns, respectively lines E45 of the current block. More specifically, it proposes to replace the application of the sub-transform A 1 or B 2 by the application of at least one partial sub-transform to a sub-vector of a column or line vector. The sub-vector obtained is returned to the column vector or output line to form the vector transformed by step E42 or E45. It is not recombined with other data from other potential transformations of the column vector or input line.
    In a known manner, the first step E42 of sub-transformation by the transform Bi can be carried out using a matrix Q of size NxN.
    In the particular example of FIG. 5 and according to the invention, the first step E42 of transformation under transform Bi is carried out using two partial sub-transforms R and S.
    For example, the step of transforming the M transposed line vectors of size N LnT [m] by the particular sub-transform Bi of size NxN is carried out by applying a first sub-transform R of size KxK to a sub-transform. set of pixels of the column vectors of the block and a second sub-transform S of size (NK) x (NK) to the rest of the pixels of the column vectors of this block. All the elements of a vector are thus treated once by only one of the two sub-partial transforms.
    From a mathematical point of view, the transform Bi can then be expressed in the form of a matrix Mbi comprising in its diagonal a sub-matrix MR of size KxK corresponding to the first sub-transform R and a sub-matrix Ms of size (NK ) x (NK) corresponding to the second sub-transform S, the rest of the elements of the matrix MBi being zero, as shown in Figure 6A.
    Thus, according to this embodiment of the invention, to transform a vector with the matrix Mbi, it suffices to perform a transformation by the matrix Mr, which processes the first K values of a vector LnT [m] and a transformation by the Ms matrix that covers the remaining NK pixels of this vector.
    In relation to FIG. 5, the transformation sub-step of an mth transposed line vector LnT [m] of the block x to be transformed by the two partial sub-transforms R and S according to the invention is detailed. A first subunit LnTi [m] of size K is formed at E421 from the first K elements of the vector Vm and a second subfrayer LnT2 [m] of size NK is formed from the remaining NK elements of the vector LnT [ m]. In E422, each sub-vector is transformed by the matrix MR, respectively Ms corresponding to the partial sub-transform R respectively S. The transformed subvectors S LnTi [m] and LnT ^ [m] m are replaced in E423 at the initial positions of the elements of the sub-vectors LnT [m] and LnT [m] in the transformed vector LnT [m].
    These operations are repeated for the M transposed line vectors of the block to be transformed. The M transformed vectors are returned to their initial positions to form the intermediate block BI.
    We will now show that splitting the matrix MBi, in a part Mr treating K points and a Ms part dealing with N-K points (with K <N-K, without loss of generality), generates a lowering of complexity.
    For a transformation of size N, applied in matrix form: • we have N * N multiplications and N * (N-1) additions.
    For a transformation of size K, applied in matrix form: • we have K * K multiplications and K * (K-1) additions.
    For a size transformation N-K, applied in matrix form: • we have (N-K) * (N-K) multiplications and (N-K) * (N-K-1) additions.
    If we cumulate the computations associated with the transformations N-K and K, we count: 1. K * K + (N-K) * (N-K) multiplications 2. K * (K-1) + (N-K) * (N-K-1) additions
    1. gives N * N + 2 * K * (K -N), where K <N so 2 * K * (KN) is negative, so the quantity is less than N * N which is the number of multiplications for the size N 2. Similarly, it is shown that the number of additions is lower than for size N.
    Note that in the particular case of a Type II DCT (refer to Table 1 above), Q is of size 8 and requires 40 operations in total, whereas the implementation of the transform of the partial transform Size 7 involves only 38 operations.
    In the particular case of type 1 DCT of size 3 and 4, it is understood that it is notoriously more favorable to process a size 3 transform which requires 6 operations to be compared to the 12 operations required by size 4.
    In a first particular example, N = 8, K = 4 and the complexity is evaluated if R and S are chosen as DCT II type transformations, for which a fast algorithm is available.
    The computational complexity is then 12 + 12 = 24 operations for the 8 pixels constituting the vector, ie 3 operations per pixel. This is to be compared to a DCT II size 8 transform, which requires 5 operations per pixel.
    In a second particular example, N = 8, K = 4 and the complexity is evaluated if R and S are chosen as transformations implemented in the form of a matrix product without an available fast algorithm.
    The computational complexity is then 28 + 28 = 56 operations for the 8 pixels constituting the vector, ie 7 operations per pixel. This is compared to a size 8 transform, which requires 15 operations per pixel.
    In a third particular example, N = 8, K = 4 and the complexity is evaluated if R is chosen as a Type II DCT and S as an applied transform in the form of a matrix product. This makes it possible to apply a suitable transformation to a part of the vectors of the block, this adapted transform can for example be of RDOT type as described in the O. Sezer article already cited.
    The computational complexity is then 12 + 28 = 40 operations for the 8 constituent pixels of the vector, ie 5 operations per pixel. This is to be compared to a size transform 8 applied as a matrix product that requires 15 operations per pixel.
    As a result, the invention notoriously reduces computational complexity. In the two cases presented above, 3/5 = 60% and 7/15 = 47% and 5/15 = 33% respectively of the complexity of a full size transformation are used.
    According to a second embodiment of the invention, the first sub-transform of size KxK is an identity-type transformation. It is expressed as a matrix containing only identical terms on its diagonal. The corresponding matrix MBi in FIG. 6B is represented for a 4x4 size block. By way of illustration, the following are the identity transformations I3 of size 3x3, I2 of size 2x2 and Ii of size 1:
    In FIG. 6C, another example of a matrix MBi is schematized, a first I2 identity transform subprocess 2 of size 2 followed by a second sub-transform of size 2x2.
    In both cases, the application of the sub-transform R = Ii respectively I2 does not require any calculation in this particular case, thus the application of the matrix MBi has the same complexity as the matrix MS.
    Three cases are presented with R = I1 and S respectively taking the form of a matrix of size N-K = 3, a DCT-II of size N-K = 3 and a DCT-II of size N-K = 7.
    Table 2
    Table 2 shows that with this embodiment, the complexity ratio is therefore 15/28 = 53% (28 corresponds to the complexity due to a 4x4 matrix of any type) and 8/12 = 67% (12 corresponds to in case one has a fast DCT II transform of size 4) and 38/40 = 95% of the initial complexity.
    The first embodiment therefore allows a gain in complexity greater than the latter mode.
    We now describe a third embodiment of the invention which is part of a context of competition of transformations. In relation with FIG. 3, the coding method identifies at E3 a transform to be applied to the current block among a plurality of predetermined transforms and, following the steps E4 to E8 of coding the current block with the aid of the identified transform, it is stores the coded data in a memory Ml. It then repeats the steps E3 to E8 encoding the current block with the other transforms in the list. In E9, the best transform is selected, based on a rate-distortion criterion.
    Arrufat et al, "Low Complexity Transform Competition for HEVC," published in the ICASSP Proceedings of March 2016, describes a method for selecting a restricted group of transforms from a list of several transforms. Separable "complete" types, optimized for distortion or based on discrete trigonometric transformations, called DTT, to which the DCT and DST transforms belong.
    In a first step, according to the prior art, the compression performance of an encoder with complete transformations, that is to say of sizes equal to those of the processed block, is evaluated. To do this, we have a set of 16 types of trigonometric transforms of size 4, for example those for which we have a fast implementation as described in the Püschel article already cited.
    For example, the list of the following transforms (8 types of DCT and 8 types of DST) is evaluated: DCT I, DCT-II, DCT-ΠΙ, DCT-IV, DCT-V, DCT-VI, DCT-VII, DCT-DCT. VIII, DST-I, DST-II, DST-ΠΙ, DST-IV, DST-V, DST-VI, DST-VII, DST-VIII.
    In total there are therefore 16x16 = 256 possible combinations when combining the vertical (Bi) and horizontal (Ai) transforms.
    To evaluate these different combinations of transforms, we measure for each pair, a compactness in a distortion / parsimony plane, as described in the article already cited by Sezer, on a set of residual image signals the following quantity:
    (3)
    Xi are the residual blocks collected on a large set of varied images. In this case, they come from blocks of size 4x4 for an intra-angular prediction of index 26 in accordance with the HEVC standard. X ^ are residual blocks quantified by thresholding resulting from the transformation of xt by Ti composed of Al and Bl. II. 11o represents the zero norm, that is the number of non-zero coefficients in ^. N, M are the dimensions of the block and H is the number of blocks considered. λ is a Lagrange multiplier that allows to set the constraint on the number of non-zero coefficients in X ^,., according to the publication of Sezer cited above.
    For example, the best transformations A 1, B 1, that is, the ones that produce the 5 weakest measures of j (1) are chosen. By learning, in accordance with Arrufat's publication, one selects the 5 most efficient transform pairs for encoding a set of residual blocks from a vertical prediction.
    In this example, at the end of the learning phase, the set of the 5 best pairs of transforms is that of the table 3:
    Table 3
    The compactness measure (equation (1)) gives a value of 29.76 for this combination of 5 pairs.
    According to the third embodiment of the invention, combinations of partial sub-transformations of sizes KxK and (N-K) x (N-K) are added to the initial set of transforms considered for learning. In the example of a 4x4 block, these are partial subprocesses of sizes lx1 and 3x3 or 2x2 and 2x2. The sub-transformations in this embodiment are of identity type. They are denoted IDi and ID2 for respective sizes 1 and 2.
    These combinations are advantageously proposed in association with a complete transform. In other words, at least two partial sub-transformations are decomposed, ie the transform A 1 or the transform B 1.
    Learning is performed from the set of transforms thus completed and, similarly to the previous case, the pairs A 1, Bi which select the least compactness measure are selected.
    We obtain the following selection of transformations:
    Table 4
    A compactness measurement of 29.34 is obtained for this new combination, which is equivalent to an improvement of 1.38% in compactness compared to the state-of-the-art embodiment. The invention is therefore remarkable in that it offers, in addition to a reduction in the number of operations as shown above, in relation to the first two embodiments, an improvement in the compression of the signal.
    Note that in the presented combinations, at least one of the partial sub-transforms is equal to the identity and the other to a DST or DCT. Of course, the invention is not limited to these examples and covers the case of combinations of at least two more complex partial sub-transforms, such as for example an ID and an adapted RDOT transform or a DST VII and an adapted RDOT transform. .
    The encoding method produces a coded data stream representative of the at least one input image. The bit stream TB produced by the coding method according to the invention is for example transmitted in the form of a signal to a decoder or a receiver comprising a decoding device via a telecommunication network.
    It is assumed that the bit stream TB has been received by a decoding device implementing the decoding method according to the invention. This decoding method will now be described in relation with FIG. 7.
    In DO, we first select as the current block C 'the first block to be processed. For example, this is the first block (in lexicographic order). This block has Nb pixels, for example 64x64 as specified in the HEVC standard for CTU blocks.
    As described for the encoding method, the current block considered in the following may be the block C 'itself or be derived from a partition of the block C' in sub-blocks CU. It can also be a residue block obtained by subtracting a prediction of a sub-block CU or the current block. We will also designate by current block a sub-block resulting from a division of a sub-block CU, residue or not, before its transformation by a transform T |.
    During a step D1, the coded data relating to the current block C 'are read. The coded data includes coding parameters, such as, for example, the prediction mode used, or in a context of transform competition, a signaling of an identifier of the transform applied to the current block and the values relating to the amplitudes and signs of the quantized residual coefficients of the current block.
    When the determined prediction mode indicates that a prediction has been made by the encoder, the current block is predicted at D2, according to the prediction mode determined from an already processed block. A predicted block Pr 'is obtained.
    During a step D3, the coded data representative of the quantized residual values of the current block (values and signs of the coefficients) are decoded, and a one-dimensional vector of values RQ '[i] with an integer of between 0 and i is formed. MxN-1. It is understood that this is the inverse operation of that of entropy encoding previously described in relation to the encoding method.
    In D4, the data of the current block RQ '[i] are dequantized. We obtain a vector R '[i],
    In D5, a reorganization of the data of the one-dimensional residual vector in the current block is performed, in a process that is the inverse of the current block path described in step E5 of FIG. 4.
    In D6, the transform to be applied to the current block is identified. In known manner, and for example in accordance with the specifications of the HEVC standard, the decoder can access the identifier ID-TR of this transform, which has been previously associated with the prediction mode of the current block. More particularly, in a context of competition of transforms, the identifier of the transform can be received in the bitstream as a coding parameter, read during step D1 and then decoded. Whether in a context of competition of transformations or not, this transform identifier actually makes it possible to identify the two orthogonal or quasi-orthogonal transforms Ai and Bi, which are successively applied to the current block.
    According to the invention, at least one of the two transformation sub-steps is implemented by applying at least two partial sub-transforms to disjoint subsets of elements of the column vectors respectively lines of the current block R ' the step of identifying the transform therefore consists in obtaining an ordered list of at least three transform identifiers, two consecutive identifiers of which correspond to the partial sub-transforms of one of the transformation sub-steps and one identifier to the transform implemented in the other substep (Ai or Bi). For example, for the second row of table 4, the identifier obtained will include .Ai = ID1, DST-IV and Bi = DCT-V.
    In a step D7, the dequantized data is applied to the transforms (Ai, Bi) corresponding to the identifiers obtained at D6. This transformation corresponds to the operation opposite to that performed at the encoder. For example, if at encoding, the transform Bi has been decomposed into two partial sub-transforms R and S, as represented in relation with FIG. 6A, we begin by applying the transpose transformation AiT of size MxM to the N column vectors. transposed from the current residue block r 'to obtain an intermediate residue block ri', then the two transposed partial subprocesses of R and S constituting the transposed transformation B | T are applied to each of the M transposed row vectors of the block ri ', the sub-transformed R being applied to the first K elements and sub-transformed S to the next NK elements of each column vector. Sub-step D721 for forming sub-vectors, D722 for applying transposed partial transforms and D743 for transforming vector composition are shown in FIG. 5 already described for the encoding method. At the end of the inverse transformation step, a signal or block r 'in the spatial domain is then obtained.
    Since the sub-transforms applied to the decoder are transposed from those applied to the encoder, the complexity generated is the same and therefore has an advantage over the state of the art.
    In a step D8, the block of pixels c 'of the decoded image is reconstructed from the block r' obtained and integrated with the image Id during decoding. If the block is a residue block, it adds the prediction Pr 'of the current block obtained during the step D2.
    During a step D9, it comes to test whether the current block is the last block to process the decoder, given the order of travel defined above. If so, the decoding process has finished processing. If not, the next step is the step of selecting the next block DO and the decoding steps DI to D9 previously described are repeated for the next block selected.
    It will be noted that the invention which has just been described can be implemented by means of software and / or hardware components. In this context, the terms "module" and "entity", used in this document, may correspond either to a software component, or to a hardware component, or to a set of hardware and / or software components, capable of implementing perform the function (s) described for the module or entity concerned.
    With reference to FIG. 8, an example of a simplified structure of a device 100 for encoding a digital image according to the invention is now presented. The device 100 implements the coding method according to the invention which has just been described in relation with FIG.
    For example, the device 100 comprises a processing unit 110, equipped with a processor μι, and driven by a computer program Pgi 120, stored in a memory 130 and implementing the method according to the invention. At initialization, the code instructions of the computer program Pgi 120 are for example loaded into a RAM memory before being executed by the processor of the processing unit 110. The processor of the processing unit 110 sets implement the steps of the method described above, according to the instructions of the computer program 120.
    In this embodiment of the invention, the device 100 comprises at least one transformer (TRANS) of the current block in a transformed block, said current block comprising M line vectors and N column vectors. The transformer is adapted to apply a first sub-transform of the N transposed line vectors for providing an intermediate residue block formed from the transformed transposed line vectors and a second sub-transformation of the M transposed line vectors of the intermediate block into transformed transposed line vectors. from which the transformed block is formed. The device 100 further comprises an encoder ENC encoder of the transformed block and a bit stream INSERT constructor comprising coded data representative of the coded image that is capable of inserting the coded coefficients.
    According to the invention, for at least one of said sub-transformations, the transformer comprises the subunits for forming at least a first sub-vector of size K <N or N respectively from adjacent elements of the input vector transforming the first sub-vector into a first transformed sub-vector by applying a partial sub-transform of size KxK and composition of the transformed vector by insertion of the at least one first transformed subvector.
    Advantageously, such a device 100 can be integrated with a user terminal TU. The device 100 is then arranged to cooperate at least with the following module of the terminal TU: a storage memory Ml, able to store in particular the intermediate coded data, in particular in a context of competition of transformations, and a module E / R of transmission / reception of data, through which the bit stream TB or the compressed file FC is transmitted in a telecommunications network, for example a wired network or a radio network.
    In relation to FIG. 9, an example of a simplified structure of a device 200 for decoding a digital image according to the invention is now presented. The device 200 implements the decoding method according to the invention which has just been described in relation with FIG. 7.
    For example, the device 200 comprises a processing unit 210, equipped with a processor μ2, and driven by a computer program Pg2 220, stored in a memory 230 and implementing the method according to the invention. At initialization, the code instructions of the computer program Pg2 220 are for example loaded into a RAM before being executed by the processor of the processing unit 210. The processor of the processing unit 210 sets implement the steps of the method described above, according to the instructions of the computer program 120.
    In this embodiment of the invention, the device comprises a decoder (DEC) of the coefficients of the current block transformed from coded data read in the bit stream, a reverse transformer (TRAINS-1) of the current block transformed into a block decoded, capable of performing two successive inverse sub-transformations, the first sub-transformation producing an intermediate block and applying to the row or row vectors of the current block, the second producing a block of pixels applying to the row vectors respectively columns of the intermediate block, resulting from the first sub-transformation.
    According to the invention, the inverse transformer comprises the subunits for forming at least a first sub-vector of size K <N or N respectively from adjacent elements of the input vector, of transformation of the first sub-vector in a first sub-vector transformed by applying a partial sub-transform of size KxK and composition of the transformed vector by insertion of the at least one first transformed subvector.
    The decoding device 200 further comprises an image constructor (RECONST) decoded from the decoded block.
    Advantageously, such a device 200 may be integrated with a user terminal TU. The device 200 is then arranged to cooperate at least with the following module of the terminal TU: a data transmission / reception module E / R, through which the bit stream TB or the compressed file FC is received from a telecommunications network, for example a wired network or a radio network.
    It goes without saying that the embodiments which have been described above have been given for purely indicative and non-limiting purposes, and that many modifications can easily be made by those skilled in the art without departing from the scope. of the invention.
    权利要求:
    Claims (12)
    [1" id="c-fr-0001]
    A method of decoding at least one digital image (Ij), from a bit stream (TB) comprising coded data representative of said image, said image (Ij) being divided into a plurality of processed blocks in a defined order, said method comprising the following steps, implemented for a block, called transformed current block (C '): decoding (D1, D3) of the coefficients of the current block from coded data read in the bitstream; Transformation (D7) of the current block into a transformed decoded block, said current block comprising M row vectors and N column vectors, with M and N non-harmonic integers, said step implementing a first substep intended to produce an intermediate block , applying to the column vectors respectively lines of the current block, the second substep intended to produce a block of pixels applying to the line vectors respectively columns of the intermediate block, resulting from the first substep; - Reconstruction (D8) of the image from the transformed decoded block; characterized in that: at least one of said first and second transformation substeps comprises, for a said line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively M from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a first partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed subvector.
    [2" id="c-fr-0002]
    2. Method for decoding at least one digital image according to claim 1, characterized in that said at least one transformation sub-step further comprises the formation of at least one second sub-vector of size less than or equal to IM -K respectively MK from adjacent elements of the input vector, not included in the first sub-vector, such that the sum of the sizes of the sub-vectors formed is equal to the size of the input vector, the application of a second partial sub-transform to said second sub-vector and in that the composition of the transformed vector comprises inserting the elements of the first and at least one second transformed sub-vector at the initial positions of the first and the at least one second sub-vector. vectors formed in the input vector.
    [3" id="c-fr-0003]
    3. Method for decoding at least one digital image according to claim 2, characterized in that one of the at least two partial sub-transforms is an identity transform.
    [4" id="c-fr-0004]
    4. Method for decoding at least one digital image according to one of claims 1 to 3, characterized in that at least one of the subvectors of the input vector is of odd size.
    [5" id="c-fr-0005]
    5. A method for decoding at least one digital image according to one of claims 2 to 4, comprising a step of selecting according to a rate-distortion criterion of a pair of transforms among a plurality of predetermined pairs, characterized in that at least one sub-transform of a said pair is decomposed into at least the first and second sub-transformed.
    [6" id="c-fr-0006]
    6. A device (200) for decoding at least one digital image, from a bit stream (TB) comprising coded data representative of said image, said image (Ij) being divided into a plurality of blocks processed in a defined order, said device comprising the following units, capable of being implemented for a block, called transformed current block (C '): decoder (DEC) of the coefficients of the current block transformed from coded data read in the bit stream ; - Transformer (TRANS1) of the current block into a decoded block, said current block comprising M row vectors and N column vectors, with M and N integers not damaged, said transformer being able to implement a first sub-unit of transformation suitable for producing an intermediate block, applying to the column vectors respectively lines of the current block, a second transforming subunit capable of producing a block of pixels applying to the row vectors respectively columns of the intermediate block, resulting from the first subunit ; - Reconstructor (RCONST) of the image from the decoded block; Characterized in that: at least one of said first and second transformation subunits comprises, for a said line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively M from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a first partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed subvector.
    [7" id="c-fr-0007]
    7. A method of coding at least one digital image, said image (Ij) being divided into a plurality of blocks of pixels processed in a defined order, said method comprising the following steps, implemented for a current block (x) , of predetermined dimensions: - Transformation (E4) of the current block into a transformed block, said current block comprising M row vectors and N column vectors, with M and N integers not damaged, said step comprising a first substep (E42) of transforming the M row vectors, respectively column, intended to provide an intermediate block formed from the row respectively transformed column vectors and a second substep (E45) for transforming the M column vectors, respectively rows of the intermediate block; - Encoding the transformed block intended to produce coded data representative of the transformed block; - Insertion of coded data in a bitstream representative of the coded picture; characterized in that at least one of said first and second transformation substeps comprises, for a line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively M from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a first partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed subvector.
    [8" id="c-fr-0008]
    8. Device for coding at least one digital image, said image (Ij) being divided into a plurality of blocks of pixels processed in a defined order, said device comprising the following units, which can be implemented for a current block (x), of predetermined dimensions: - Transformer (E4) of the current block into a transformed block, said current block comprising M row vectors and N column vectors, with M and IM not damaged, said transformer comprising a first sub-unit of transforming the M row vectors, respectively column, able to provide an intermediate block formed from the row respectively transformed column vectors and a second transformation subunit of the M column vectors, respectively rows of the intermediate block; - Converted block encoder for producing coded data representative of the transformed block; - Manufacturer of a binary train representative of the encoded image adapted to insert said coded data in the bit stream; characterized in that at least one of said first and second transformation subunits comprises, for a line vector respectively column, said input vector: the formation of at least a first sub-vector of size K <N respectively M from adjacent elements of the input vector; transforming the first sub-vector into a first transformed sub-vector by applying a first partial sub-transform of size KxK; and the composition of the vector transformed by insertion of the at least one first transformed subvector.
    [9" id="c-fr-0009]
    9. Signal carrying a bit stream (TB) comprising coded data representative of at least one digital image (Ij), said digital image being divided into blocks of processed pixels in a defined order, a transformed block being obtained by transformation of the pixels of a current block, said current block including M row vectors and N column vectors, with M and N integers not damaged, said transformation comprising a first substep intended to produce an intermediate block, applying to the row vectors respectively columns of current block, the second one intended to produce a block of pixels applying to the column vectors respectively lines of the intermediate block, resulting from the first substep, characterized in that, at least one of said first and second transformation substeps comprising the formation of a sub-vector of size K <N respectively M from adjacent elements of undit line vector respectively colon ne, the application of a first partial sub-transform of size KxK to said sub-vector and the composition of a line vector respectively column transformed by insertion of the transformed sub-vector, said signal comprises an identifier representative of said sub-vector; partial transform.
    [10" id="c-fr-0010]
    10. User terminal (TU) characterized in that it comprises a device for decoding at least one digital image according to claim 6 and a device for coding at least one digital image according to claim 8.
    [11" id="c-fr-0011]
    11. A computer program comprising instructions for implementing the method of decoding at least one digital image according to one of claims 1 to 5, when executed by a processor.
    [12" id="c-fr-0012]
    A computer program comprising instructions for implementing the method of encoding at least one digital image according to claim 7, when executed by a processor.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP0294357B1|1994-09-14|Picture signals coding method
    EP3183879A1|2017-06-28|Method for encoding a digital image, and associated decoding method, devices and computer programmes
    WO2017187040A1|2017-11-02|Method of decoding a digital image, method of coding, devices, and computer programs associated therewith
    EP3254467A1|2017-12-13|Method for encoding a digital image and associated decoding method, devices and computer programs
    Siddeq et al.2015|A novel 2D image compression algorithm based on two levels DWT and DCT transforms with enhanced minimize-matrix-size algorithm for high resolution structured light 3D surface reconstruction
    EP3075155B1|2019-02-06|Method for encoding and decoding images, device for encoding and decoding images and corresponding computer programs
    WO2016102804A1|2016-06-30|Method for encoding a digital image and associated decoding method, devices and computer programs
    WO2017037368A2|2017-03-09|Method of coding and decoding images, device for coding and decoding images and computer programmes corresponding thereto
    KR101982788B1|2019-05-27|Method and apparatus for encoding and decoding video signal using improved prediction filter
    Naidu et al.2014|A novel framework for JPEG image compression using baseline coding with parallel process
    FR3044507A1|2017-06-02|IMAGE ENCODING AND DECODING METHOD, IMAGE ENCODING AND DECODING DEVICE AND CORRESPONDING COMPUTER PROGRAMS
    FR2956789A1|2011-08-26|METHOD AND DEVICE FOR PROCESSING A VIDEO SEQUENCE
    FR3057130B1|2019-08-02|METHOD OF ENCODING AN IMAGE, DECODING METHOD, DEVICES, TERMINAL EQUIPMENT AND COMPUTER PROGRAMS
    FR2927745A1|2009-08-21|Digital signal filtering method for telecommunication system, involves determining optimal filter based on criterion that depends on values of sub-signal, and associating optimal filter with context function corresponding to sub-signal
    EP3314897A1|2018-05-02|Method for encoding a digital image, and associated decoding method, devices and computer programs
    FR3046321A1|2017-06-30|DIGITAL IMAGE ENCODING METHOD, DECODING METHOD, DEVICES, USER TERMINAL, AND ASSOCIATED COMPUTER PROGRAMS
    WO2019057368A1|2019-03-28|Methods for encoding and decoding an image
    FR3050858A1|2017-11-03|IMAGE ENCODING AND DECODING METHOD, IMAGE ENCODING AND DECODING DEVICE AND CORRESPONDING COMPUTER PROGRAMS
    RU2429541C2|2011-09-20|Method of encoding digital images using discrete wavelet transformation of adaptively defined basis
    US20210314615A1|2021-10-07|Data processing apparatuses, methods, computer programs and computer-readable media
    FR3083042A1|2019-12-27|IMAGE DECODING METHOD, CODING METHOD, DEVICES, TERMINAL EQUIPMENT AND COMPUTER PROGRAMS THEREOF
    EP3854088A1|2021-07-28|Methods and devices for encoding and decoding a data stream representing at least one image
    FR3057429A1|2018-04-13|METHOD FOR ENCODING A DIGITAL IMAGE, DECODING METHOD, DEVICES AND COMPUTER PROGRAMS
    FR2918203A1|2009-01-02|METHODS AND APPARATUSES FOR ENCODING AND DECODING DIGITAL IMAGES BASED ON THE TYPE OF SOURCE BLOCKS, SIGNAL, AND CORRESPONDING COMPUTER PROGRAM PRODUCTS.
    FR3058019A1|2018-04-27|METHOD FOR ENCODING AND DECODING IMAGE PARAMETERS, DEVICE FOR ENCODING AND DECODING IMAGE PARAMETERS AND CORRESPONDING COMPUTER PROGRAMS
    同族专利:
    公开号 | 公开日
    US10771817B2|2020-09-08|
    US20190158878A1|2019-05-23|
    JP6948346B2|2021-10-13|
    WO2017187040A1|2017-11-02|
    FR3050598B1|2020-11-06|
    KR20180136481A|2018-12-24|
    JP2019515552A|2019-06-06|
    EP3449631A1|2019-03-06|
    CN109196861A|2019-01-11|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    CN1290338C|1997-11-14|2006-12-13|索尼电子有限公司|1/4 size real time decoding of digital video|
    WO2010090484A2|2009-02-09|2010-08-12|삼성전자 주식회사|Video encoding method and apparatus using low-complexity frequency transformation, and video decoding method and apparatus|
    US9075757B2|2009-06-24|2015-07-07|Qualcomm Incorporated|16-point transform for media data coding|
    US20120127003A1|2009-08-06|2012-05-24|Youji Shibahara|Coding method, decoding method, coding apparatus, and decoding apparatus|
    JP5259879B2|2010-07-15|2013-08-07|株式会社東芝|Image encoding method and image decoding method|
    EP2624561A2|2010-09-28|2013-08-07|Samsung Electronics Co., Ltd|Method and device for the transformation and method and device for the reverse transformation of images|
    EP2727355A1|2011-06-29|2014-05-07|Motorola Mobility LLC|Methods and system for using a scan coding pattern during intra coding|FR3044507A1|2015-11-30|2017-06-02|Orange|IMAGE ENCODING AND DECODING METHOD, IMAGE ENCODING AND DECODING DEVICE AND CORRESPONDING COMPUTER PROGRAMS|
    WO2021054798A1|2019-09-21|2021-03-25|엘지전자 주식회사|Transform-based image coding method and device|
    CN112346967B|2020-10-20|2022-03-01|四川长虹电器股份有限公司|Pc-lint cloud service system based on cloud platform, computer equipment and storage medium|
    法律状态:
    2017-04-19| PLFP| Fee payment|Year of fee payment: 2 |
    2017-10-27| PLSC| Publication of the preliminary search report|Effective date: 20171027 |
    2018-04-18| PLFP| Fee payment|Year of fee payment: 3 |
    2019-04-17| PLFP| Fee payment|Year of fee payment: 4 |
    2020-04-27| PLFP| Fee payment|Year of fee payment: 5 |
    2021-04-06| PLFP| Fee payment|Year of fee payment: 6 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1653704A|FR3050598B1|2016-04-26|2016-04-26|PROCESS FOR DECODING A DIGITAL IMAGE, PROCESS FOR CODING, DEVICES, AND ASSOCIATED COMPUTER PROGRAMS|FR1653704A| FR3050598B1|2016-04-26|2016-04-26|PROCESS FOR DECODING A DIGITAL IMAGE, PROCESS FOR CODING, DEVICES, AND ASSOCIATED COMPUTER PROGRAMS|
    KR1020187032786A| KR20180136481A|2016-04-26|2017-04-05|A method, a coding method, devices, and associated computer programs for decoding a digital image|
    EP17719662.3A| EP3449631A1|2016-04-26|2017-04-05|Method of decoding a digital image, method of coding, devices, and computer programs associated therewith|
    JP2018555984A| JP6948346B2|2016-04-26|2017-04-05|Digital image decoding methods, coding methods, equipment and accompanying computer programs|
    PCT/FR2017/050817| WO2017187040A1|2016-04-26|2017-04-05|Method of decoding a digital image, method of coding, devices, and computer programs associated therewith|
    US16/092,359| US10771817B2|2016-04-26|2017-04-05|Method for decoding a digital image, coding method, devices, and associated computer programs|
    CN201780026361.5A| CN109196861A|2016-04-26|2017-04-05|Method, coding method, equipment and the related computer program of decoding digital image|
    [返回顶部]