FR2766034A1 - Compressed data coding-decoding method - Google Patents

Abstract

The coding-decoding method operates on a set of data representing physical quantities, the set being comprised of sub-sets fn. For each sub-set it is divided into blocks of data to be coded, a reference set is divided into source blocks and then, for each block of data to be coded, a predetermined number of source blocks in the reference sub-set are selected. The method then determines a transformation between the source blocks selected and the block to be coded.

Description

 The present invention relates, in general, to a method and a device for encoding data, as well as to a method and a device for decoding data. Data encoding and decoding aims in particular to compress and decompress data.

 Compressed data can be stored in a reduced memory space compared to the uncompressed initial data, or can be transmitted faster, or at a lower rate, than the uncompressed initial data.

 These data are a priori arbitrary and represent physical quantities. In preferred applications, image coding and decoding are more particularly considered, since images are known to occupy a large memory space and require a high bit rate, or a significant transmission time, when it is desired to transmit them.

 Many methods of image coding currently exist.

Among these, the progressive coding methods code the data progressively, for example by successive iterations, and the coding precision increases as the coding iterations increases.

 Progressive data coding allows progressive transmission of coded data. In the case of an image, it is possible to construct during the transmission of the image an approximate version of the transmitted image. The approximate version is constructed from the data already received without waiting for all the data relating to the image to have been transmitted. The received image is gradually improved as the data is transmitted.

 Progressive transmission allows a user to quickly view the approximate version of the image. The user can interrupt the transmission in progress if the approximate version of the image is sufficient or on the contrary if the image is not the one desired.

 For example, JPEG coding (Joint Picture Expert Group) is a block coding method based on discrete cosine transformations, followed by compression by quantization and variable length coding. JPEG coding can be progressive according to a specific mode called HJPEG (Hierarchic Joint Picture Expert Group).

 The HJPEG coding is relatively long and complex to implement, because it is based on transformation calculations, in this case transformation in discrete cosine. Likewise, decoding requires calculations of the same complexity and length.

 The present invention aims to remedy the drawbacks of the prior art, by providing a data coding method and device, as well as the corresponding decoding method and device, which allow simpler and therefore faster coding and decoding, while having a data compression rate at least equivalent to the rates of the prior art.

To this end, the invention proposes a method of coding a set of data representative of physical quantities,
characterized in that it comprises, the step of:
- division of the set into data blocks to be coded,
then the following steps, performed iteratively a predetermined number of times:
- construction of a reference set,
- division of the reference set into source blocks,
and, for each of the data blocks to be coded:
- selection of a predetermined number of source blocks in said reference set,
- determination of a transformation between the selected source blocks and the block to be coded.

Correlatively, the invention proposes a device for coding a set of data representative of physical quantities,
characterized in that it comprises:
means for dividing the set into data blocks to be coded,
- reference assembly construction means,
- means for dividing the reference set into source blocks,
means for selecting a predetermined number of source blocks in each of the reference sets, for each of the data blocks to be coded, and
means for determining a transformation between the selected source blocks and the block to be coded.

 The invention relates to the field of progressive coding and decoding with loss of data, in particular digital images. Coding is carried out by block.

 The invention allows rapid compression and decompression of data. The invention is more particularly intended for the transmission of compressed data. In this context, the method also includes the step of transmitting the transformation determined for each of the data blocks.

 According to a preferred characteristic, it is sought, for each of the data blocks to be coded if a mapping of the block to be coded with one of the source blocks respects a matching criterion. The block is then coded by means of a transformation only if the criterion is not satisfied.

Block matching coding is fast and requires few bits. Decoding is also very fast.

 The matching criterion of the matching step comprises a measurement of a first distance between the block to be coded and the source block. This type of criterion is simple to implement.

 According to another preferred characteristic, a set of data to be coded is constructed at each coding iteration, prior to the coding proper.

 At the first iteration, this set is constructed by subtracting an average of the set of data representative of physical quantities from said set of data representative of physical quantities.

 From the second iteration, this set is constructed by subtracting the set to be coded in the previous iteration from a set of data reconstructed from the transformations determined in the previous iteration.

 Thus, the set of data to be coded at each of the iterations results from a difference. For example in the case of an image, a difference image is generally coded with greater precision than the original image.

 According to a preferred characteristic, the transformation minimizes a second distance between the block to be coded and its approximation calculated by applying the transformation to the selected source blocks. This distance is a difference, for example the square error, between the values of the data of the block to be coded and the values of its approximation calculated by applying the transformation to the source blocks. This type of distance is simple and quick to calculate.

 According to another preferred characteristic, the transformation is a multilinear approximation, which allows fast and precise coding. For specific applications, the transformation can be a multilinear approximation combined with a geometric transformation, or a polynomial on the values of the coefficients of the source blocks.

 According to characteristics of the invention, for the first coding iteration of the set to be coded, said at least one reference set is a predetermined set, and, from the second iteration, said at least one reference set is the set which was previously coded and then decoded in the previous iteration. The reference set is thus determined in a simple manner.

 To increase the accuracy and speed of coding, the source blocks are selected from a part of the reference set which depends on the part in which the block to be coded is located in the set being coded.

 To simplify the implementation, the source blocks have a size which is determined as a function of the size of the block to be coded, for example the source blocks have a size multiple of a factor F of that of the block to be coded and are sub- sampled by the factor F, where F is an integer, or the source blocks have the same size as the block to be coded.

 According to other characteristics, the block to be coded is of predetermined size, which is simple to implement, or the size of the block to be coded decreases as iterations progress, which increases the coding precision at each iteration. . Indeed, the blocks of variable size allow coding which adapts to the content of the image, and consequently a more precise coding of the latter.

In another aspect, the invention proposes a method for decoding a set of coded data representative of physical quantities,
The data set being coded by a series of coding sets which each include coded representations of data blocks, each of the blocks being coded by means of a transformation between selected source blocks in a reference set and the block ,
characterized in that it comprises, for each of the coding sets successively, and for each of the blocks, the step of applying the transformation to the source blocks to decode the block.

Correlatively, the invention proposes a device for decoding a set of coded data representative of physical quantities,
The set being coded by a series of coding sets which each include coded representations of data blocks of the data set, each of the blocks being coded by means of a transformation between source blocks selected from a set of reference and block,
characterized in that it comprises means for applying the transformation to the source blocks, for each of the coding sets successively and for each of the blocks, for reconstructing the block.

 The method and the decoding device make it possible to obtain advantages analogous to those of the method and the coding device.

 According to a preferred characteristic, the method includes a search step to determine whether the block is coded by matching blocks, prior to decoding by applying a transformation.

 In the event of a positive response to the search step, the block is decoded by copying a source block determined by the matching of blocks. This decoding is therefore particularly simple and rapid.

 In the case where the coding has been carried out on sets constructed by differences, the decoding method further comprises, for the first iteration, the addition of an average to the set resulting from the application of the transformation to source block for each of the blocks, and, from the second iteration, the addition of the set resulting from the decoding to the previous iteration and of the set resulting from the application of the transformation to the source block for each of the blocks.

The characteristics and advantages of the present invention will appear more clearly on reading a preferred embodiment illustrated by the attached drawings, in which:
FIG. 1 schematically represents a data coding device according to the invention,
FIG. 2 schematically represents a device for decoding data according to the invention,
FIG. 3 represents an embodiment of the data coding device according to the invention,
FIG. 4 represents an embodiment of the data decoding device according to the invention,
FIG. 5 is a data coding algorithm according to a first embodiment of the invention,
FIG. 6 schematically represents iterations of coding according to the invention,
FIG. 7 is an image divided into blocks during its coding according to the invention,
FIG. 8 is an image divided into blocks during its coding according to the invention,
- Figures 9a to 9f show blocks to be coded and associated source blocks
FIG. 10 is a data decoding algorithm according to the first embodiment of the invention,
FIG. 11 schematically represents iterations of decoding according to the invention,
FIG. 12 is a data coding algorithm according to a second embodiment of the invention,
FIG. 13 schematically represents iterations of coding according to the second embodiment of the invention,
FIG. 14 schematically represents iterations of decoding according to the second embodiment of the invention,
FIG. 15 is a data coding algorithm according to a third embodiment of the invention,
FIG. 16 is a data decoding algorithm according to the third embodiment of the invention,
FIG. 17 is a data coding algorithm according to a fourth embodiment of the invention,
FIG. 18 schematically represents a step in the algorithm of FIG. 17,
FIG. 19 schematically represents iterations of coding according to the fourth embodiment of the invention,
FIG. 20 is a data decoding algorithm according to the fourth embodiment of the invention, and
- Figure 21 shows schematically iterations of decoding according to the fourth embodiment of the invention.

 According to an embodiment chosen and shown in FIG. 1, a data coding device 2 according to the invention comprises an input 22 to which is connected a source 1 of non-coded data. The source 1 comprises for example a memory means, such as random access memory, hard disk, floppy disk, compact disc, for memorizing non-coded data, this memory means being associated with a suitable reading means for reading the data there. Means for storing the data in the memory means can also be provided.

 The source 1 can for example be a digital camera in the case of a signal representative of a still image, or a digital camera delivering an image sequence. In the latter case, each image in the sequence can be coded as a still image, independently of the other images in the sequence, or the sequence is coded as a three-dimensional signal (two spatial dimensions and one temporal dimension).

 Means 3 users of coded data are connected at output 27 of the coding device 2.

 The user means 3 comprise for example means for storing coded data, and / or means for transmitting the coded data.

 FIG. 2 represents a device 5 for decoding data coded by the device 2.

 Means 4 users of coded data are connected at the input 52 of the decoding device 5. The means 4 comprise for example coded data memory means, and / or means for receiving the coded data which are adapted to receive the coded data transmitted by the transmission means 3.

 Means 6 users of decoded data are connected to the output 57 of the decoding device 5. The user means 6 are for example means for viewing images.

 The coding device and the decoding device can be integrated into the same digital camera, for example a digital photographic camera provided with a display screen.

 An embodiment of the data coding device 2 is shown in FIG. 3.

The coding device 2 comprises a data and address bus 20 to which are connected:
an input buffer memory 21, also connected to input 22 of the device, for temporarily storing the data to be coded,
a read only memory (ROM) 23 in which is stored a coding algorithm which will be explained below,
a controller 24 for implementing the coding algorithm,
a random access memory (RAM) 25 for storing in registers parameters modified during the execution of the coding algorithm, and
- an output buffer memory 26, also connected to output 27 of the device.

 An embodiment of a data decoding device 5 is shown in FIG. 4.

the decoding device 5 comprises a data and address bus 50 to which are connected:
an input buffer memory 51, also connected to input 52 of the device, for temporarily storing the data to be coded,
a read only memory (ROM) 53 in which is stored a decoding algorithm which will be explained below,
a controller 54 for implementing the decoding algorithm,
a random access memory (RAM) 55 for storing in registers parameters modified during the execution of the decoding algorithm, and
- an output buffer memory 56, also connected to output 57 of the device.

 The operation of the coding 2 and decoding 5 devices will be detailed during the description of the corresponding algorithms.

According to the embodiment chosen and represented in FIGS. 5 and 6,
The invention applies to the coding of a set of data representative of physical quantities which is a digital image f, here a gray level image, comprising for example 128 × 96 pixels at 256 levels.

 The algorithm represented in FIG. 5 comprises steps E1 to E15 which carry out the coding of the digital image. Overall, as shown in FIG. 6, the image is coded progressively, according to several coding iterations. The coding result of each iteration can be transmitted, so that the coded image is transmitted gradually, with a limited bit rate.

 Step E1 is an initialization to which the source 1 supplies an image to be coded f to the coding device 2. An integer M, explained below, is determined as a function of the size of the image. A number of coding iterations N is also determined. The parameters M and N can be predetermined or adjusted by a user.

 The next step E2 is a division of the image considered f into the number M of blocks A1 to AM of adjacent pixels. In this embodiment, the blocks A1 to AM are adjacent, without overlap. In a variant not shown, the blocks can overlap, to limit the block effects on decoding, to the detriment of the compression rate.

 According to a first variant of this embodiment shown in FIG. 7, the blocks all have the same predetermined size, and are for example square blocks of 32 × 32 pixels in the case of the 128 × 96 pixel image. The image division is uniform throughout the image. According to a second variant represented in FIG. 8, the size of the blocks, while being predetermined, depends on the location of the blocks in the image: in a determined area of the image, for example a central area capable of containing more details, the blocks are smaller than at the edges of the image.

This determined area will then be coded more finely than the rest of the image.

 Two different block sizes are shown in Figure 8 32x32 and 16x16 pixels. However, a higher number of sizes can be envisaged for specific applications, in particular if it is known that the quantity of information varies greatly from one zone to another of the image to be coded.

 In all cases, the division of the image results in a set of M blocks A1 to AM, stored in memory 25. The blocks A1 to AM will be compressed, or coded, one after the other, in order any. This order is preferably predetermined and is then the order in which the blocks will then be decompressed, or decoded, which makes it unnecessary to locate the blocks by an index or by coordinates. As a variant, it is possible to locate the blocks and to associate the identification, index or coordinates, with the coded block, which makes it possible to decompress the blocks in any order, which may be different from the order used for compression.

 Step E2 is followed by step E3, which is an initialization of a parameter n to one, to start the first iteration of coding. Step E3 is followed by step E4 in which an additional image g0 is considered as a reference image for the first coding iteration. The additional image g0 is arbitrary, for example a uniformly gray image.

 Step E4 is followed by step E5 which is an initialization of a parameter m to 1, to consider the first block A1 in image f.

 The next step E6 is the division of a reference image into blocks of adjacent pixels, stored in memory 25. The size of the blocks depends on the size of the blocks to be coded Arn. Preferably, the blocks of the reference images are F2 times larger than the blocks to be coded, where F is an integer for example equal to two, which makes it possible to exploit many details in the reference image. The blocks of the reference images are then sub-sampled by the factor F, to work on blocks having the same number of pixels as the blocks to be coded. According to a variant that is simpler to implement, the blocks of the reference images have the same size as the blocks to be coded. According to a preferred embodiment which is more particularly taken into account here, the reference image g.1 is the coded and then decoded image according to the invention during the iteration preceding the coding iteration in progress, except for the first iteration, for which the reference image g0 is predetermined.

Step E6 is followed by step E7 which is the selection of source blocks Dni, rn, i to Dn-1.m, K in the reference image relative to the block to be coded
Image arnd f.

 The number K of source blocks is predetermined, and is for example equal to 4. As a variant, the number K is a parameter for adjusting the algorithm. The source blocks are selected by their location in the reference image g1 as a function of the location of the block to be coded Arn in the image f.

In general, the source blocks are selected in a region of the reference image which depends on the region of the block to be coded in the image considered. Preferably, the two regions are similar, that is to say that the superposition of the image considered and of the reference image leads to partial overlapping of the block to be coded and of the source blocks.

 Figures 9a to 9f show different examples of source block selection. In each of FIGS. 9a to 9f, the block to be coded Arn of the image considered has been superimposed on the source blocks Dn 1, m, 1 to Dn1, rnA of the source image gon 1. In the six examples, four are considered source blocks of the same size as the Arn coding block.

 In FIG. 9a, the block to be coded Arn is in a central part of the image f; the four source blocks are adjacent and the block to be coded is generally centered on all of the four selected source blocks.

 In FIGS. 9b to 9e, the block to code Arn is located on an edge of the image represented by a line in dotted lines. Two of the source blocks partially overlap the other two source blocks, so that the block to be coded covers part of each of the source blocks.

 In FIG. 9f, the block to be coded Arn is in a corner of the image, and the four source blocks partially overlap, so that the block to be coded covers part of each of the source blocks.

 As a variant, when the block to be coded is on an edge or in a corner of the image, the four source blocks are adjacent, and two or three of the source blocks are truncated and completed either by a uniformly gray complement or by a complement determined by extension of the image for example by symmetry or by periodization of the image.

 Step E7 is followed by step E8, at which a transformation between the source blocks and the block to be coded is calculated.

 According to a preferred embodiment, the transformation is a multilinear approximation, of the form: Tn, m = {anm, k, bnm} making it possible to construct the block ARn, m = #k (an, mn, kx Dn-1, mk) + bn, m.

In this expression, an, m, k is a scale function, here a constant function, or coefficient, for a given source block Dn-1, m, k and of norm less than unity, and bn, m is a gray level shift function, here a constant function, or coefficient, for the block to be coded Arn at iteration n. The function an mk applies to each of the pixels of the block Dn-1, m, kt, which here amounts to multiplying all the pixels by the coefficient an m k-
The functions, here the coefficients, an, m, k and bn, rn are determined to minimize a distance d (Am, ARn, m) between the block to be coded Arn and the reconstructed block ARnm at iteration n by applying the transformation Tn, m to the source blocks according to the formula given above.

 According to an embodiment which is simple to implement, the functions n, rn, k and bn, m are discretized and a limited number of possible values for the coefficients n, rn.k is tested. For example, each coefficient an, m, k can take two integer values, a positive value and the opposite value. All combinations of coefficients an, m, k are tested, and for each combination, the coefficient bn, m is chosen to minimize the distance considered.

 The distance is calculated between the gray levels of the block to be coded Arn and the gray levels of the block ARn, rn resulting from the application of the approximation to the source blocks Dn-1, m, 1 to Dn 1, m, K .

 The distance can be the absolute value of the sum of the differences, or the mean square error, or the absolute value of the largest difference, calculated between the source block Arn and the reconstructed block by applying the transformation Tn, rn to the blocks source.

 In a preferred embodiment, the distance is the mean square error calculated between the gray levels of the block to be coded Arn and the gray levels of the block ARn, m resulting from the application of the transformation to the source blocks Dn-1 , m, 1 a Dn-1, m, K.

 The combination Tn, m = n, rn, k, bn, m} which minimizes the distance is stored in step E9 in the memory 25 of the coding device 2 (FIG. 3) as a compressed form of the block Arn. The set of compressed blocks {Tn, rn} relative to the image f and determined at the iteration n forms a compressed image hn.

According to other embodiments
- the transformation combines a multilinear approximation calculated on the source blocks and geometric transformations, such as rotation, of the source blocks,
the transformation is of polynomial form, for example of order two or three, calculated on the values of the pixels of the source blocks. This type of transformation gives more precise results, while taking longer to calculate.

 The next step E10 is a test to check if m is equal to M, that is to say if the block which has just been coded is the last of the image f. If at least one block remains to be coded in the image f, the step E10 is followed by the step E11 in which the parameter m is incremented by one, to consider the next block in the image f. Step EI is followed by step E7 previously described.

When the block which has just been coded is the last of the image f,
The step E10 is followed by the step E12 which is the transmission of the compressed image hn. This transmission is a step of the progressive transmission of the coded image and will allow the decoding device to decode an approximation of the image, as discussed below.

The next step E13 is a test to check whether all the iterations of coding of the image f have been carried out. If this is not the case, step E13 is followed by step E14 which is the construction of the decompressed image 9n
The application of each of the transformations Tn, m previously determined to the source blocks reconstructs each of the blocks ARnm, the whole of which forms the decoded image gn. The image gn is stored in memory 25 to be used as a reference image for the following iteration n + 1.

 Step E14 is followed by step E15 to increment the parameter n by one. Step E15 is followed by step E5 previously described.

In the case where n = N, that is to say for the last iteration of coding of the image f, it is not necessary here to reconstruct the image gN.

 When the response to step E13 is positive, all the iterations for coding all the blocks of the image to be coded. Coding is thus faster, since the selection of the source blocks is carried out only once per iteration; the distortions introduced by the coding will however be greater than with the embodiment described above.

 With reference to FIGS. 10 and 11, the first embodiment of the algorithm for progressive decoding of compressed digital image comprises steps E20 to E30. Step E20 is an initialization at which the parameter n is initialized to 1, for the first decoding iteration.

 In the next step E21, an additional image g0 is considered as a source image for the first decoding iteration. The additional image g0 is identical to that considered in step E4 previously described, and is for example a uniformly gray image.

 From the next step E22, the compressed form hn of the image is considered. Step E22 is the reading of the compressed form hnl and the initialization of the parameter m to 1, to consider the first block to be reconstructed. The blocks are preferably treated in the same order as in compression, which avoids having to memorize and process an index, or coordinates, to locate the blocks.

 The following step E23 is the division of the reference image g1 into blocks of adjacent pixels, which are the source blocks stored in memory 55.

Step E23 is identical to step E6. The next step E24 is the selection of source blocks Dn-1, m, 1 to Dn 1 m K in the reference image gn-1, relative to the block to be reconstructed ARnrn of the image to be decompressed hn The image of reference is selected in the same way as for coding. According to the preferred embodiment, the reference image gn-1 is the decoded image preceding the image hn. The source blocks are selected in the same way as for coding, with the same possible variants.

Step E24 is followed by step E25 in which the transformation Tn, rn defined by the parameters an mk and bn, rn is applied to the source blocks Dn 1, m, 1 to Dn-1, m, K selected at l 'previous step E24, according to the formula:
ARn, m = zk (an, m, k X Dn-l, m, k) + bn, m
Step E25 results in the decompressed block ARn, m stored in memory 55.

 The next step E26 is a test to determine whether the block which has just been decompressed is the last of the image being decompressed. If this is not the case, step E26 is followed by step E27 in which the parameter m is incremented by one unit to consider the next block. Step E27 is followed by step E24 previously described.

When all the blocks of the current image have been decompressed,
Step E26 is followed by step E28 in which the decompressed image gn is reconstructed and for example displayed on a screen, and / or stored. In general, Step E28 is the transfer of the image gn to the user means 6.

The next step E29 is a test to determine whether all the iterations have been carried out. If there is at least one iteration to complete, Step
E29 is followed by step E30 in which the parameter n is incremented by one, to go to the next iteration. Step E30 is followed by step E22 previously described.

 Thus, the image g1 is first of all reconstructed, then the image g2, and so on, as shown in FIG. 11, which allows a progressive construction of the decoded image, each iteration making it possible to obtain a image with more details than in the previous iteration.

 Another embodiment of the coding method according to the invention is represented in FIGS. 12 and 13 in the form of an algorithm comprising steps E31 to E45 stored in memory 23 (FIG. 3).

 In this embodiment, the size of the blocks to be coded decreases, and the number of blocks to be coded increases correspondingly, at each coding time, so as to code increasingly fine details.

Steps E31 to E45 are respectively analogous to steps
E1 to E15 described with reference to FIG. 5 The differences compared to the previous embodiment are as follows:
- The initialization step E31 includes the initialization of the parameter n to one, corresponding to the first coding iteration.

The step E32 of dividing the image fa for the result of the blocks, the size tn of which decreases, and the number increases, as and when successive iterations,
- Step E33 is a test to determine if the parameter n is equal to one, corresponding to the first iteration if the response is positive, Step E33 is followed by step E34, otherwise, Step E33 is followed by l 'step E35,
- Step E45 is followed by step E32 of dividing the image f.

 As shown in FIG. 13, the aim of the first coding iteration is to calculate a first set of transformations {T1, m}, each of these transformations corresponding to the coding of a block of the image f from source blocks of image g0. The transformations are applied to the source blocks of image 90 to determine the image g1.

The purpose of the second coding iteration is to calculate a second set of transformations {T2m}, each of these transformations corresponding to the coding of a block of the image f from source blocks of the image g1. The blocks of the image f have, for this iteration, a size smaller than that of the blocks considered in the previous iteration. Correlatively, the source blocks also have a size smaller than those of the previous iteration. The transformations are applied to the source blocks of image g1 to determine the image g2
The third iteration uses the image g2 as the reference image, and a set of transformations T3, rn} is determined between blocks of image f and source blocks of image g2, the size of the blocks to be coded and reference blocks being decreased from the previous iteration. This iteration makes it possible to calculate the image g3.

 The decoding algorithm corresponding to this embodiment is identical to that represented in FIG. 10. FIG. 14 represents the successive iterations of decoding. At the first iteration of decoding, the first set of transformations i, rn} is applied to the respective blocks, having the first predetermined size t1, of the image g0. Image g1 is thus reconstructed.

 The image g1 is divided into blocks having the second predetermined size t2, then the second set of transformations is applied to the blocks of the image g1, to form the image g2, and so on.

 The images reconstructed as iterations contain increasingly fine details.

 Another embodiment of the coding method according to the invention is shown in FIG. 15, and includes steps E51 to E691 stored in memory 23 (FIG. 3). Compared to the first embodiment, this embodiment consists in firstly finding, for each block to be coded, whether it can be coded by matching with the block of the reference image which has the same size. and which occupies the same position in the image. If this coding is possible, the block is coded by block matching otherwise, the block is coded by means of a determined transformation as in the first embodiment.

 Steps E51 to E56 are identical to steps E1 to E6 previously described with reference to the first embodiment (FIG. 5).

 Step E56 is followed by step E57 which is a test for determining whether the block to be coded current Arn has been coded by block mapping to the previous iteration. This test is carried out on an indicator In 1 mX for example a bit set to the value one if the block Am is coded by block matching at 'iteration n-l.

 If the answer is positive, step E58 is followed by step E59, in which an indicator In m is stored in step E59, to indicate that the block considered is coded by block matching. This indicator is for example a bit set to the value one. The indicator In, m represents the coding information, at iteration n, of the block Arn.

 If the answer is negative in step E58, the latter is followed by step E60 in which the current block to be coded Arn is coded by block matching: it is sought in the reference image gn-1 on ARn block 1 m of the same size and having the same position in the image.

 The next step E61 is a test to check whether a criterion is met. This criterion here consists in verifying whether a distance calculated between the block to be coded Arn and the correspondence block ARn, m is less than a predetermined threshold. If the response is positive, an indicator Inm is stored in step E59, to indicate that the block considered is coded by block matching. This indicator is for example a bit set to the value one. The indicator In, m represents the coding information, at iteration n, of the block Arn.

 Step E59 is followed by steps E65 to E691, respectively identical to steps E10 to E15 previously described.

If the answer is negative in step E61, that is to say if the coding by block matching does not give a satisfactory result, The step
E61 is followed by steps E62 to E691, respectively identical to steps
E7 to E15 previously described. Step E63 results in a transformation Tn, m. . The indicator Inm is set to zero and is associated with each transformation Tn, m.

 As a variant, the two preceding embodiments are combined, first of all we try to code the blocks by matching blocks, and then, if this first coding is not possible, by means of a transformation; the coding steps by determining a transformation then include varying the size of the block to be coded as iterations progress. In this case, steps E57 and E58 are useless, since the size of the blocks varies from one iteration to the next. Step E56 is then followed directly by step E60 of block matching.

 The decoding method corresponding to the coding in FIG. 15 is shown in FIG. 16 and includes steps E70 to E91 stored in memory 53 of the device 5 (FIG. 4).

Steps E70 to E73 are respectively identical to steps
E20 to E23 described with reference to Figure 9.

 The next step E74 is a test to determine whether the block under reconstruction ARnm has been coded by matching of blocks. The test consists in analyzing the value of the indicator In, m and in searching if the indicator Inm is at value one.

 If the response is positive, then the following step E75 consists in reading from the reference image gn n-1 the correspondence block ARn1m and in copying this block as a decoded block ARnm. The reconstruction of the ARnm block is therefore very fast, since it is a block copy. Step E75 is followed by steps E78 to E91 which are respectively identical to steps E25 to E30 previously described (FIG. 9).

 If the response to step E74 is negative, step E74 is followed by steps E76 to E91 which are respectively identical to steps E24 to E30 previously described (FIG. 9).

 Another embodiment of the coding method according to the invention is shown in FIGS. 17 to 19, and includes steps E101 to E115 stored in memory 23 (FIG. 3). Compared to the first embodiment, this embodiment consists in working at each iteration on image differences and in coding only the additional details compared to the previous iteration.

 The steps E101 and E102 are identical to the steps E1 and E2 previously described (FIG. 5).

 The next step E103 is the calculation of the average value M1 to MM of each of the blocks A1 to AM of the image f, and the subtraction, pixel by pixel, of this average value Mrn to each of the blocks Arn. Step E103 results in an image f1, as shown in FIG. 18. The image z0 of the mean values has a compressed form ho comprising the mean value M1 to MM of each of the blocks Arn. The compressed form ho is stored and / or transmitted for use during decoding.

 Step E103 is followed by step E104 which is the division into blocks A 1 to Al.M of the image f '.

 As a variant, the steps E103 and E104 are deleted, and one works directly on the image f, as in the previous embodiments.

 The next step E105 is an initialization of a parameter n to one, to start the first iteration of coding. An additional image g0 is considered as a reference image for the first coding iteration. The additional image g0 is arbitrary, for example a uniformly black image in the case where we are working on image differences.

An image f, is equal to the image f1.

 Step E105 is followed by step E106 which is an initialization of a parameter m to 1, to consider the first block An 1 in the image to be coded at iteration n, ie the image f'n.

 The next step E107 is the division of the reference image gn-1 into blocks of adjacent pixels, stored in memory 25.

 Step E107 is followed by step E108 which is the selection of source blocks Dn.î, rn, î to Dn-1, m, K in the reference image relative to the block to be coded An, rn of l image to code f'n.

 Step E108 is followed by step E109, at which a transformation Tn, m between the selected source blocks and the block to be coded is calculated. The transformation Tn, m which minimizes the distance d (Anm, ARn, m) is stored in the memory 25 of the coding device 2 (FIG. 3) as a compressed form of the block Ans. The set of compressed blocks {Tn, m} determined at iteration n forms a compressed image hn.

 The next step E110 is a test to check if m is equal to M, that is to say if the block which has just been coded is the last of the current image f'n.

If at least one block remains to be coded in the image fn, the step E110 is followed by the step El II in which the parameter m is incremented by one, to consider the next block in the image f ' not. The El Il stage is followed by the stage
E107 previously described.

When the block which has just been coded is the last of the image f'n,
Step E110 is followed by step E112 which is the transmission of the compressed image hn. This transmission is a step in the progressive transmission of the coded image and will allow the decoding device to decode an approximation of the image, as explained below.

The following step Eu 13 is a test to check whether all the iterations of coding of the image f have been carried out. If this is not the case, step E113 is followed by step Eu 14 which is the construction of the decompressed image gn
The application of each of the transformations Tn, m previously determined to the source blocks reconstructs each of the blocks ARnm, the whole of which forms the decoded image gn. The image gn is stored in memory 25 to be used as a reference image for the following iteration n + 1. The image fn + 1 to be coded in the next iteration n + 1 is determined by subtracting the image f'n from the image gn.

 Step Eu 14 is followed by step E115 to divide the image f'n + 1 into blocks to be coded An + 1 1 to An + 1, M. These blocks will be coded in the next iteration.

The next step E116 increments the parameter n by one unit to go to the next iteration. Step E116 is followed by step E106 previously described. In the case where n = N, that is to say for the last iteration of coding of the image f, it is not necessary here to reconstruct the image gN
When the response to step Ex 13 is positive, all the iterations of coding of the image f have been carried out.

 In this embodiment, the different aspects of selecting source blocks and determining transformations are identical to those described for the previous embodiments.

 The decoding method corresponding to the coding in FIG. 17 is shown in FIGS. 20 and 21 and includes steps E201 to E211 stored in memory 53 of the device 5 (FIG. 4).

 Step E201 is an initialization at which the parameter n is initialized to 1, for the first decoding iteration.

 In the next step E202, an additional image g0 is considered as a source image for the first decoding iteration.

The additional image g0 is identical to that considered in step E105 previously described, and is for example a uniformly black image.

The compressed average image ho is considered.

 The next step E203 is the reading of the compressed file hn, and the initialization of the parameter m to 1, to consider the first block to be reconstructed. The blocks are preferably treated in the same order as in compression, which avoids having to memorize and process an index, or coordinates, to locate the blocks.

 The next step E204 is the division of the reference image g.1 into blocks of adjacent pixels, which are the source blocks stored in memory 55.

Step E204 is identical to step E107. The next step E205 is the selection of source blocks Dni, rn, i to Dn-1, m, K in the reference image gn-1, relative to the block to be reconstructed ARnrn of the image to be decompressed gn.

The reference picture is determined in the same way as for coding.

The source blocks are selected in the same way as for coding, with the same possible variants.

Step E205 is followed by step E206 in which the transformation Tn, rn defined by the parameters an mk and bn, m is applied to the source blocks Dn- 1, m, 1 to Dn1, rn, K selected at previous step E205, according to the formula: ARn, rn ((an, m, k X Dn-1, m, k) + bn, m
Step E206 results in the decompressed block ARnm stored in memory 55.

 The next step E207 is a test to determine whether the block which has just been decompressed is the last of the image being decompressed. If this is not the case, step E207 is followed by step E208 in which the parameter m is incremented by one unit to consider the next block. Step E208 is followed by step E205 previously described.

When all the blocks of the current image have been decompressed,
Step E207 is followed by step E209 in which the decompressed image gn is reconstructed from the blocks ARn, m. The approximation Zn of the image at iteration n is determined by adding the image gn to the approximation Zn-1 determined in the previous iteration. The approximation z, is for example displayed on a screen, and / or stored. In general, step E209 is the transfer of the image z, to the user means 6.

The next step E210 is a test to determine whether all the iterations have been carried out. If at least one iteration remains, the step
E210 is followed by step E211 in which the parameter n is incremented by one, to go to the next iteration. Step E21 1 is followed by step
E203 previously described.

 Thus, the image z1 is first of all reconstructed, then the image z2, and so on, as shown in FIG. 21, which allows a progressive construction of the decoded image, each iteration making it possible to obtain a image with more details than in the previous iteration.

 The invention has been described for a grayscale image, that is to say an image with a single component; it also applies to color images conventionally comprising three components.

 These components are conventionally the red, green and blue components, or the components of luminance and chrominance. Each of these components is treated as exposed for a single component image.

 As stated above, the invention applies to an image sequence. In this case, each image in the sequence can be coded as a still image, independently of the other images in the sequence, or the sequence is coded as a three-dimensional signal (two spatial dimensions and one temporal dimension).

 Of course, the present invention is not limited to the embodiments described and shown, but encompasses, quite the contrary, any variant within the reach of ordinary skill in the art.

Claims (38)

 1. Method for coding a set (f) of data representative of physical quantities,  characterized in that it comprises, the step of:  - division (E2) of the set into data blocks to be coded (Am)  then the following steps, performed iteratively a predetermined number of times:  - construction (E4, E14) of a reference set (gn-1),  - division (E6) of the reference set (gn-1) into source blocks,  and, for each of the data blocks to be coded:  selection (E7) of a predetermined number of source blocks (Dn-1, m, 1,, 1, m, K) in said reference set,  - determination (E8) of a transformation (Tnm) between the selected source blocks and the block to be coded.  2. Coding method according to claim 1, characterized in that it further comprises the step of transmitting the transformation determined for each of the data blocks.  3. Method for coding a set (f) of data representative of physical quantities,  characterized in that it includes the step of:  - division (E52) of the set into data blocks to be coded,  then the following steps, performed iteratively a predetermined number of times:  - construction (E54, E66) of a reference set (g, .l),  - division (E56) of the reference set (gn-1) into source blocks,  and, for each of the data blocks to be coded:  - matching (E57) of the block to be coded with one of the source blocks, and, if a matching criterion is not satisfied (E58),  - selection (E60) of a predetermined number of source blocks (Dn 1, m, 1, Dn 1, m, K) in said reference set,  - determination (E61) of a transformation between the selected source blocks and the block to be coded.  4. Coding method according to claim 3, characterized in that it further comprises the step of transmitting, for each of the data blocks, either of the indication that the block is coded by block matching, or of the transformation determined for the data block.  5. Coding method according to claim 3 or 4, characterized in that the correspondence criterion of the matching step comprises a measurement of a first distance between the block to be coded (Anm) and the source block (Dn 1, m, j).  6. Method for coding a set (f) of data representative of physical quantities,  characterized in that it comprises the following steps, carried out iteratively a predetermined number of times:  - construction (E103, E114) of a set of data to be coded (f'n).  - determination (E109) of a transformation (Tnm) between the selected source blocks and the block to be coded.  - selection (E108) of a predetermined number of source blocks (Dni, rn, Dn-l, m, K) in said reference set,  and, for each of the data blocks to be coded:  - division (E107) of the reference set (gn-1) into source blocks,  - construction (E105, El 14) of a reference set (gn-1),  - division (E104, E115) of the set into data blocks to be coded (An, m),  7. Coding method according to claim 6, characterized in that the step of constructing a set of data to be coded comprises, at the first iteration, subtracting an average from the set of data representative of physical quantities said set of data representative of physical quantities.  8. Coding method according to claim 6, characterized in that the step of constructing a set of data to be coded comprises, from the second iteration, the subtraction of the set to be coded from the previous iteration from a set of data reconstructed from the transformations determined in the previous iteration.  9. Coding method according to any one of claims 1 to 8, characterized in that the transformation minimizes a second distance (d (Am, ARnrn), d (An, m, ARnrn)) between the block to be coded (Am , An m) and its approximation calculated by applying the transformation to the selected source blocks (Dn 1, m, 1, Dn-1, m, K).  10. Coding method according to any one of claims 1 to 9, characterized in that the transformation is a multilinear approximation.  11. Coding method according to any one of claims 1 to 9, characterized in that the transformation is a multilinear approximation combined with a geometric transformation.  12. Coding method according to any one of claims 1 to 9, characterized in that the transformation is a polynomial on the values of the coefficients of the source blocks.  13. Coding method according to any one of claims 6 to 12, characterized in that the calculation of the second distance comprises the calculation of a difference between the values of the data of the block to be coded (Am, An, m) and the values of its approximation (ARnm) calculated by applying the transformation to the source blocks.  14. Coding method according to any one of claims 6 to 12, characterized in that the second distance is the mean square error calculated between the values of the block to be coded (Am, m) and the values of the block resulting from l of the approximation to the source blocks (Dn.i, rn.i to Dn- 1, mK) -  15. Coding method according to any one of claims 1 to 14, characterized in that, for the first coding iteration of the set to be coded, said at least one reference set (g0) is a predetermined set.  16. Coding method according to any one of claims 1 to 15, characterized in that, from the second iteration, said at least one reference set is the set (gn-1) which has been previously coded and then decoded on the previous iteration.  17. Coding method according to any one of claims 1 to 16, characterized in that the source blocks (Dn-1, m, 1, Dn.1, rn, K) are selected from a part of the set of reference (gn-1) which depends on the part in which the block to be coded (Am) is located in the assembly being coded (f).  18. Coding method according to any one of claims 1 to 17, characterized in that the source blocks (Dn-1, m, k) have a size which is determined as a function of the size of the block to be coded (Am)  19. Coding method according to any one of claims 1 to 18, characterized in that the source blocks (Dni.rn, i, 1, m, K) have a size multiple of a factor F2 of that of the block to code (Am) and are subsampled by the factor F, where F is an integer.  20. Coding method according to any one of claims 1 to 18, characterized in that the source blocks (Dni, rn.i, Dn-1, m, K) have the same size as the block to be coded (Am)  21. Coding method according to any one of claims 1 to 20, characterized in that the block to be coded (Am) is of predetermined size.  22. Coding method according to any one of claims 1 to 20, characterized in that the size of the block to be coded (Am) decreases as iterations progress.  23. Method for decoding a set of coded data representative of physical quantities, the data set being coded by a series of coding sets (hn) which each include coded representations of data blocks (Tn, m) , each of the blocks being coded by means of a transformation (Tn, m) between source blocks selected from a reference set and the block,  characterized in that it comprises, for each of the coding sets successively, and for each of the blocks, the step of applying (E25) the transformation to the source blocks to decode the block (ARnm).  24. Method for decoding a set of coded data representative of physical quantities, the data set being coded by a series of coding sets (hn) which each include coded representations of data blocks (Tn, rn) ,  characterized in that it comprises, for each of the coding sets successively, and for each of the blocks, the steps of:  - search (E74) if the block is coded by matching of blocks,  and, in the event of a negative response, the block then being coded by means of a transformation (Tn, m) between source blocks selected in a reference set and the block:  - application (E77) of the transformation to the source blocks to decode the block.  25. A decoding method according to claim 24, characterized in that, in the event of a positive response to the search step, the block is decoded (E75) by copying a source block determined by the matching of blocks.  26. A decoding method according to any one of claims 23 to 25, characterized in that it also comprises, for the first iteration, the addition of an average (ho) to the set resulting from the application transformation to the source blocks for each of the blocks.  27. A decoding method according to any one of claims 23 to 26, characterized in that it also comprises, from the second iteration, the addition of the set resulting from the decoding to the previous iteration and of the set resulting from the application of the transformation to the source block for each of the blocks.  28. Coding device (2) for a set (f) of data representative of physical quantities,  characterized in that it comprises (24):  means for dividing the set (f) into data blocks to be coded (Am),  - means of reference set construction (9n.i)  - means for dividing the reference set (gn-1) into source blocks,  means for selecting a predetermined number of source blocks (Dn 1, m, 1, Dn 1, m, K) in each of the reference sets (gn-1), for each of the data blocks to be coded (Am ) and  means for determining a transformation (tam) between the selected source blocks and the block to be coded.  29. Coding device (2) according to claim 28, characterized in that it comprises means (3) for transmitting the transformation determined for each of the blocks to be coded.  30. Coding device (2) for a set (f) of data representative of physical quantities,  characterized in that it comprises (24):  means for dividing the set into data blocks to be coded (Am),  - reference assembly construction means (9n-i),  - means for dividing the reference set (gn-1) into source blocks,  means for matching, for each of the data blocks to be coded, the block to be coded with one of the source blocks, and, if a correspondence criterion is not satisfied,  means for selecting a predetermined number of source blocks (Dn-1, m, 1, Dn-1, m, K) in each of the reference sets, for each of the data blocks, and  means for determining a transformation (Tn, m) between the selected source blocks and the block to be coded  31. Coding device (2) according to claim 30, characterized in that it comprises transmission means (3), for each of the blocks to be coded, either of the indication that the block is coded by matching of block, or of the transformation determined for the data block.  32. Coding device (2) according to any one of claims 28 to 31, characterized in that it further comprises means for constructing a data set to be coded (f,).  33. Coding device according to any one of claims 28 to 32, characterized in that the means of division, construction, selection and determination are incorporated in  - a microcontroller (24),  - a read only memory (23) comprising a program for coding each of the data blocks, and  - a random access memory (25) comprising registers adapted to record variables modified during the execution of said program.  34. Decoding device (5) of a set of coded data representative of physical quantities, the set being coded by a series of coding sets (hn) which each include coded representations of data blocks of the set of data, each of the blocks being coded by means of a transformation between source blocks selected from a reference set (go 1) and the block,  characterized in that it comprises means (54) for applying the transformation to the source blocks, for each of the coding sets successively and for each of the blocks, for reconstructing the block (ARnm).  35. Device for decoding (5) a set of data representative of physical quantities, the set being composed of subsets (hn) which each include coded representations of data blocks of the data set,  characterized in that it comprises (54):  - search means, for each of the blocks (Am) if the block is coded by matching of blocks,  and in the event of a negative response, the block then being coded by means of a transformation (Tn, m) between source blocks selected in a reference set and the block:  - means of applying the transformation to the source blocks to reconstruct the block (ARn, m).  36. Decoding device according to claim 34 or 35, characterized in that the application means are incorporated in:  - a microcontroller (54),  - a read only memory (53) comprising a program for decoding each of the data blocks, and  - A random access memory (55) comprising registers adapted to record variables modified during the execution of said program.  37. Digital apparatus comprising means suitable for implementing the coding method according to any one of claims 1 to 22, and / or means suitable for implementing the decoding method according to any one of claims 23 to 27.  38. Digital apparatus comprising the coding device according to any one of claims 28 to 33, and / or the decoding device according to any one of claims 34 to 36.