FR2862450A1 - Coding/decoding digital signals for digital images having signals block divided transferred then sub band quantized using symbol groups/construction entropic code group/code group sets - Google Patents

Abstract

The digital signal coding and decoding process divides signals in digital blocks, transforms them into sub bands and quantizes the blocks. There is a step (E8) quantizing symbol groups, followed by construction (E18) of an entropic code for each quantized group. Each entropic code group is then coded (E19).

Description

1 2862450

  The present invention generally relates to digital signal coding and decoding.

  The purpose of the coding is to compress the signal, which makes it possible to transmit or memorize, respectively, the digital signal by reducing the transmission time, or the transmission rate, respectively by reducing the memory space used.

  The invention lies in the field of lossy compression of digital signals.

  More particularly, digital image coding and decoding is considered.

  A known digital image encoding mode is JPEG encoding, for example described in W. B. Pennebaker and J. L. Mitchell, entitled JPEG: Still image data standard compression, Van Nostrand Reinhold, 1993.

  It is an object of the present invention to provide an image coding method and apparatus which provides a compression ratio over distortion which is higher than that of the prior art.

  To this end, the invention proposes a method of encoding a digital signal representative of physical quantities comprising the steps of: dividing the signal into data blocks, transforming the data blocks into frequency sub-bands; transformed data blocks for forming blocks of quantization symbols, characterized in that it comprises the steps of: 2 2862450 - formation of quantization symbol groups according to the rank of each symbol in its respective block, - construction an entropy coding mode for each group of quantization symbols, - encoding each group by its respective entropy coder.

  The invention makes it possible to use an entropic coder for all the symbols of the digital signal which are at the same rank in the different blocks formed in the signal. In addition, the entropy encoder used is constructed specifically for each group of symbols. Thus, the compression ratio is maximum.

  According to a preferred characteristic, the transformation is a DCT transformation of the data blocks. This type of transformation is commonly used, especially for digital images.

  According to a preferred characteristic, the construction of an entropy coding mode for each group comprises for a given group: calculation of the group distribution, construction of an entropy coding mode for the group as a function of its distribution.

  This makes it possible to take into account the particular statistics of the group.

  According to a preferred characteristic, the entropy coding is an arithmetic coding. This type of coding gives satisfactory experimental results while being simple to implement.

  The invention also relates to a method for decoding a digital signal encoded by the previously presented method, characterized in that it comprises the steps of: - construction of an entropy decoding mode respectively associated with each of the groups entropy decoding of the symbols of each of the groups; formation of symbol blocks according to the respective group of each symbol.

  According to one embodiment, the invention relates to a method characterized in that it comprises the steps of: - first division of the signal into data blocks, 3 2862450 - first transformation into frequency sub-bands of the data blocks, - first quantizing the transformed data blocks, to form quantization symbol blocks, - constructing an entropy encoding mode for each quantization symbol group, - dividing the signal into data blocks, - second transforming into subbands. frequency of the data blocks, - second quantization of the transformed data blocks, to form quantization symbol blocks, - coding of each group by its respective entropy coder.

  In this embodiment, the necessary memory space is reduced, since the results of the division, transformation and quantization steps are not stored.

  Correlatively, the invention relates to a device for coding a digital signal representative of physical quantities comprising: means for dividing the signal into data blocks, means for transforming the data blocks into frequency sub-bands, quantization means for the transformed data blocks, for forming blocks of quantization symbols, characterized in that it comprises: means for forming quantization symbol groups as a function of the rank of each symbol in its respective block; means for constructing an entropy encoding mode for each group of quantization symbols; means for encoding each group by its respective entropy coder.

  The invention also relates to a decoding device.

  The devices according to the invention comprise means for implementing the characteristics previously described.

  The coding device, the method and the decoding device have advantages similar to those previously presented.

  The invention also relates to a digital apparatus including the device according to the invention or means for implementing the method according to the invention.

  This digital camera is for example a digital camera, a digital camcorder, a scanner, a printer, a photocopier, a fax machine. The advantages of the device and the digital device are identical to those previously exposed.

  An information storage means, readable by a computer or by a microprocessor, integrated or not into the device, possibly removable, stores a program implementing the method according to the invention.

  A computer program readable by a microprocessor and comprising one or more sequences of instructions is able to implement the methods according to the invention.

  The features and advantages of the present invention will become more apparent upon reading a preferred embodiment illustrated by the accompanying drawings, in which: FIG. 1 shows an embodiment of a device implementing FIG. FIG. 2 represents a coding device according to the invention and a corresponding decoding device; FIG. 3 represents a first embodiment of coding method according to the invention; FIG. signal to be processed according to the invention; FIG. 5 represents an exemplary signal to be processed according to the invention; FIG. 6 represents an embodiment of a decoding method according to the invention; FIG. embodiment of coding method according to the invention.

  According to the embodiment chosen and shown in FIG. 1, a device embodying the invention is for example a microcomputer 10 connected to different peripherals, for example a digital camera 107 (or 2862450 a scanner, or any means of acquisition or image storage) connected to a graphics card and providing information to be processed according to the invention.

  The device 10 comprises a communication interface 112 connected to a network 113 able to transmit digital data to be processed or conversely to transmit data processed by the device. The device 10 also comprises a storage means 108 such as for example a hard disk. It also includes a disk drive 110 110. This disk 110 may be a diskette, a CD-ROM or a DVD-ROM, for example. The disk 110 as the disk 108 may contain data processed according to the invention as well as the program or programs implementing the invention which, once read by the device 10, will be stored in the hard disk 108. According to a variant, the program allowing the device to implement the invention, can be stored in ROM 102 (called ROM in the drawing). In the second variant, the program can be received to be stored in a manner identical to that previously described via the communication network 113.

  The device 10 is connected to a microphone 111. The data to be processed according to the invention will in this case be an audio signal.

  This same device has a screen 104 making it possible to display the data to be processed or to interface with the user who can thus set up certain modes of treatment, using the keyboard 114 or by any other means (mouse for example) .

  The CPU 100 (called the CPU in the drawing) executes the instructions relating to the implementation of the invention, instructions stored in the read-only memory 102 or in the other storage elements. At power-up, the processing programs stored in a non-volatile memory, for example the ROM 102, are transferred into the RAM RAM 103 which will then contain the executable code of the invention as well as registers for storing the variables. necessary for the implementation of the invention.

  More generally, an information storage means, readable by a computer or by a microprocessor, integrated or not into the device, possibly removable, stores a program implementing the method according to the invention.

  The communication bus 101 allows communication between the various elements included in the microcomputer 10 or connected to it. The representation of the bus 101 is not limiting and in particular the central unit 100 is able to communicate instructions to any element of the microcomputer 10 directly or through another element of the microcomputer 10.

  Referring to Figure 2, an embodiment of coding device 2 according to the invention is for coding a digital signal for the purpose of compressing it. The coding device is integrated in a device, which is for example a digital camera, a digital video camera, a scanner, a printer, a photocopier, a fax machine, a database management system or a computer.

  An image to be encoded IM is stored in a memory 1. The encoder receives the IM image.

  The coding device 2 comprises: means 20 for dividing the signal into data blocks, means 21 for transforming the data blocks into frequency sub-bands, means 22 for quantizing the transformed data blocks to form quantization symbol blocks.

  According to the invention, the coding device according to the invention further comprises: - means 23 for forming groups of quantization symbols according to the rank of each symbol in its respective block, - means 24 for constructing a entropic coding mode for each group of quantization symbols; means 25 for coding each group by its respective entropy coder.

  The operation of the coding device is detailed below using algorithms.

  The coded picture can be transmitted by a transmission module 3 whose operation is conventional to a decoding device 4. In a variant, the coded picture is simply stored for later decoding.

  The decoding device 4 comprises: means 40 for constructing an entropy decoding mode 5 respectively associated with each of the groups; means 41 for decoding entropy symbols of each of the groups; means 42 for forming blocks. of symbols, depending on the respective group of each symbol.

  The operation of the decoding device is detailed below using an algorithm.

  The decoding device 4 provides a decoded picture IM 'to an image display device 5.

  It should be noted that the coding device and the decoding device 15 can be integrated in the same device, for example the computer 10 of FIG.

  FIG. 3 represents a first embodiment of an image coding method, according to the invention. This method is implemented in the coding device and comprises steps E1 to E24.

  The method is embodied in the form of an algorithm that can be stored in whole or in part in any information storage means capable of cooperating with the microprocessor. This storage means is readable by a computer or a microprocessor. This storage means is integrated or not to the device, and can be removable. For example, it may comprise a magnetic tape, a floppy disk or a CD-ROM (compact disk with frozen memory).

  Step E1 is a division into blocks B; of the image to be coded. In general, a block is a set of samples of the image. For example, adjacent square blocks of predetermined size 8x8 pixels are formed in the image. Of course, other block sizes are possible.

  The next step E2 is an initialization to which a first block is considered.

  The next step E3 is a linear or non-linear transformation of the current block B. For example, a discrete cosine transformation, called DCT according to the English Discrete Cosine Transform, is applied to the current block. Alternatively, a discrete wavelet transformation is applied to the current block.

  The step E3 further comprises a transformation of the two-dimensional signal and a one-dimensional signal. For this, preferably, the block B; is traveled in a zig-zag order. This technique, called zig zag scan, is described in the document entitled Digital image compression techniques of M. Rabbani and P.W. Jones, in SPIE Opt. Eng. Press, Bellingham, Washington, 1991.

  As shown in FIG. 4, in block B; at 8x8 coefficients, we consider the diagonals of coefficients oriented from bottom left to top right. The route starts at the top left. The diagonals are covered either from bottom left to top right, or in the opposite direction. For a given diagonal, the direction of travel is the opposite direction to that of the previous diagonal.

  Following this direction of travel, the coefficients are arranged in a vector V; one-dimensional containing 64 coefficients.

  Alternatively, as shown in Figure 5, the block B; is traveled line by line, from left to right and from top to bottom, to form a vector V '; dimension also containing 64 coefficients.

  The next step E4 is a quantization of the current vector V. All the samples of the block are divided by a quantization step Q, then rounded to the nearest integer. The result is a list of 64 signed integer coefficients.

  The quantization step is either predetermined or selected by the user. It may be the same for all samples or different for each sample of the vector.

  The values thus obtained are transformed into quantization symbols. A non-zero value represents its own quantization symbol. When one or more coefficients are zero, they are replaced by a specific quantization symbol RX, where the integer x is the number of null consecutive coefficients.

  9 2862450 If the last coefficient obtained corresponds to a series of zeros, it is not taken into account.

  For example, let V be; the following vector: V; = [12.3 2.5 -6.0 -0.1 4.1 3, 2 -80.0 4.9 18.4 -26.8 2.3 -6.7 0.4 0.1 - 0.2 0.5] With a quantization step of 10, the quantized vector QV; is the following: QV; = [1 0 -1 0 0 0 -8 0 2 -3 0 -1 0 0 0 0] After transforming the obtained values into quantization symbols, the vector SQV is obtained; next: SQV; = [1 RI -1 R3 -8 RI 2 -3 RI -1] Alternatively, it is possible not to group the successive zeros together and to code each of them independently of the others.

  The next step E5 is an initialization in which two integer variables M and n are initialized to zero. The variable M represents the number of quantization symbols contained in the current quantized vector SQV; The variable n represents the index of a current group in which the current quantization symbol is classified.

  The next step E6 is a test to determine if the quantization symbols of the current block are all null. If the response is positive, then this step is followed by step E13 which is described below.

  If the answer is negative in step E6, then this step is followed by step E7, which is an initialization at which a first quantization symbol is considered in the current quantized vector SQV.

  The next step E8 is the assignment of the current symbol to the current symbol group G. In the next step E9, the variable n is incremented by one unit to consider the next symbol group.

  The next step E10 is the incrementation of the variable M by one unit.

  The next step E11 is a test to determine if the current quantization symbol is the last of the current quantized vector SQV; If the answer is negative, then this step is followed by step E12 at which the next quantization symbol 2862450 is considered. Step E12 is followed by step E8 previously described.

  When the response is positive in step El 1 this step is followed by step E13 which is the storage of the value of the variable M associated with the current block.

  Step E13 is followed by step E14 which is a test to determine if the current block is the last block to be processed. If the answer is negative, then this step is followed by step E15 to which a next block is considered. Step E15 is followed by step E3 previously described.

  When all the blocks have been processed, step E14 is followed by step E16 in which the first group of quantization symbols is considered.

  The next step E17 is the calculation of the distribution of the current symbol group Gn. For this, the histogram of the group is determined, that is to say the number of symbols of the group having the same value, for each possible value.

  The next step E18 is the construction of an entropy encoder for the current group. For this purpose, for example, a Huffman coder is used. The probability of occurrence of each symbol is calculated by dividing the number of occurrences of each symbol in the group by the number of symbols in the group. A code is then assigned to each symbol. The structure and length of the code depend on the likelihood of the symbol appearing.

  In a preferred embodiment, the coding used is an arithmetic coding.

  It should be noted that information intended to subsequently build the associated decoder is written in the compressed file containing the data representing the image. For example, for a Huffman coder, one writes in the compressed file either the code associated with each symbol, or the probabilities of appearance of each symbol allowing the decoder to reconstruct the codes.

  The next step E19 is the coding of the symbols of the current group Gn by the previously constructed entropy coder. The result is a bit stream that is written to the compressed file containing the encoding data of the image.

  Step E19 is followed by step E20 which is a test to determine if the current group is the last group to be processed. If the answer is negative, then this step is followed by step E21 to which a next group is considered. Step E21 is followed by step El 7 previously described.

  When all the groups have been processed, the step E20 is followed by the step E22 at which the distribution of the signal M representing the set of values of the variable M associated with the blocks is calculated. It is recalled that the variable M represents the number of quantization symbols respectively contained in each of the blocks.

  The next step E23 is the construction of an entropic coder for the signal M from its previously determined distribution. In a preferred embodiment, the entropy encoder is an arithmetic encoder. Alternatively, a Huffman encoder is used. Information for constructing the associated decoder is also written to the compressed file.

  The next step E24 is the coding of the signal M by the previously constructed entropy coder. The result is a bit stream that is written to the compressed file containing the encoding data of the image.

  Fig. 6 shows an embodiment of an image decoding method according to the invention. This method is implemented in the decoding device and comprises steps E30 to E48.

  The method is embodied in the form of an algorithm that can be stored in whole or in part in any information storage means capable of cooperating with the microprocessor. This storage means is readable by a computer or a microprocessor. This storage means is integrated or not to the device, and can be removable. For example, it may comprise a magnetic tape, a floppy disk or a CD-ROM (compact disk with frozen memory).

  The step E30 is the construction of the entropic decoder of the signal M. This construction is carried out on the basis of the information contained in the compressed file.

  The next step E31 is the entropy decoding of the symbols of the signal M. The result is stored in memory.

  The next step E32 is an initialization to which a first group of quantization symbols is considered.

  The next step E33 is the construction of the entropic decoder of the current group G. This construction is carried out on the basis of the information contained in the compressed file.

  The next step E34 is the entropy decoding of the symbols of the current group.

  Step E34 is followed by step E35 which is a test to determine if the current group is the last group to be processed. If the answer is negative, then this step is followed by step E36 to which a next group is considered. Step E36 is followed by step E33 previously described.

  When all groups have been processed, step E35 is followed by step E37 at which a first block is considered.

  The next step E38 is the selection of a first group of symbols. A variable m is initialized to the value zero. The variable m is used to count the number of decoded symbols in the current block.

  The next step E39 is the selection of a first quantization symbol in the current group.

  The next step E40 is a test to determine whether the number m of symbols already decoded in the current block is equal to the number M of symbols to be decoded in the current block. If the answer is negative, then this step is followed by step E41 at which the current symbol is decoded and inserted into the decoded block. If the decoded symbol is of the form RX, then x zeros are inserted. Otherwise, the value of the symbol is inserted.

  The next step E42 is an incrementation of the number m of one unit to signify that a symbol has been decoded.

  At the next step E43, the next group is considered.

  In the next step E44, a next quantization symbol in the current group is considered.

  Step E44 is followed by step E40 previously described.

  When the answer is positive in step E40, then all the symbols of the current block have been decoded. This step is followed by the step E45 which is the dequantization of the current block. Each previously obtained value is multiplied by the quantization step Q. The next step E46 is the transformation of the one-dimensional dequantized block into a two-dimensional block of 8 × 8 DCT coefficients. For this, an operation called zig-zag reverse scan is performed on the block obtained in the previous step.

  The thus decoded block is placed in the decoded picture.

  The next step E47 is a test to determine if the current block is the last block to be processed. If the answer is negative, then this step is followed by step E48 at which a next block is considered.

  Step E48 is followed by step E38 previously described.

  When the answer is positive in step E47, the decoding of the image is completed.

  FIG. 7 represents a second embodiment of an image coding method, according to the invention. This method is implemented in the coding device and comprises steps E101 to E131.

  The method is embodied in the form of an algorithm that can be stored in whole or in part in any information storage means capable of cooperating with the microprocessor. This storage means is readable by a computer or a microprocessor. This storage means is integrated or not to the device, and can be removable. For example, it may comprise a magnetic tape, a floppy disk or a CD-ROM (compact disk with frozen memory).

  The steps E101 to E114 forming the first part of the method are respectively similar to the steps E1 to E14 previously described, with the exception of the step E108.

  Step E101 is a division into blocks B; of the image to be coded.

  The next step E102 is an initialization to which a first block is considered.

  The next step E103 is a linear or non-linear transformation of the current block B. For example, a discrete cosine transformation, called DCT according to the English Discrete Cosine Transform, is applied to the current block. Alternatively, a discrete wavelet transformation is applied to the current block.

  Step E103 further comprises a transformation of the two-dimensional signal and a one-dimensional signal. For this, preferably, the block B; is traveled in a zig-zag order.

  The next step E104 is a quantization of the current vector V. The result is a quantized vector SQV.

  The next step E105 is an initialization in which two integer variables M and n are initialized to zero. The variable M represents the number of quantization symbols contained in the current quantized vector SQV; The variable n represents the index of a current group in which the current quantization symbol is classified.

  The next step E106 is a test to determine if the quantization symbols of the current block are all null. If the answer is positive, then this step is followed by step E113 which is described below.

  If the answer is negative at step E106, then this step is followed by step E107 which is an initialization at which a first quantization symbol is considered in the current quantized vector SQV.

  In the following step E108, a current group Gn is considered and its histogram is incremented. The histogram is the number of symbols in the group with the same value, for each possible value. The number associated with the current quantization symbol is incremented by one unit in the histogram of the current group.

  In the next step E109, the variable n is incremented by one unit to consider the next symbol group.

  The next step E110 is the incrementation of the variable M by one unit. The next step El11 is a test to determine if the symbol of

  current quantization is the last of the quantized current vector SQV; If the answer is negative, then this step is followed by step E112 at which the next quantization symbol is considered. Step E112 is followed by step E108 previously described.

  When the answer is positive in step El11 this step is followed by step E113 which is the storage of the value of the variable M associated with the current block.

  Step E113 is followed by step E114 which is a test to determine if the current block is the last block to be processed. If the answer is negative, then this step is followed by step E115 to which a next block is considered. Step E115 is followed by step E103 previously described.

  When all blocks have been processed, step E114 is followed by step E116.

  The result of the first part of the method is the histogram of each of the groups G and the number of symbols to be encoded in each of the groups. Only these values are stored. It should be noted that the quantized vectors are not memorized, which makes it possible to work with a reduced memory.

  The second part of the process comprising steps E116 to E131 is now described.

  Step E116 is the construction of an entropic encoder for each of the groups. For this purpose, for example, a Huffman coder is used. The probability of occurrence of each symbol is calculated by dividing the number of occurrences of each symbol in the group by the number of symbols in the group. A code is then assigned to each symbol. The structure and length of the code depend on the likelihood of the symbol appearing.

  In a preferred embodiment, the coding used is an arithmetic coding.

  It should be noted that information intended to subsequently build the associated decoder is written in the compressed file containing the data representing the image.

  Step E117 is a division into B blocks; of the image to be coded.

  The next step E118 is an initialization to which a first block is considered.

  The next step E119 is a linear or non-linear transformation of the current block B. For example, a discrete cosine transformation, called DCT according to the English Discrete Cosine Transform, is applied to the current block. Alternatively, a discrete wavelet transformation is applied to the current block.

  Step E119 further comprises a two-dimensional signal transformation and a one-dimensional signal. For this, preferably, the block B; is traveled in a zig-zag order.

  The next step E120 is a quantization of the current vector V; The result is a quantized vector SQV;

  The next step E121 is an initialization in which the integer variable n is initialized to zero. The variable n represents the index of a current group in which the current quantization symbol is classified.

  The next step E122 is a test to determine if the quantization symbols of the current block are all null. If the answer is positive, then this step is followed by the step E127 which is described below.

  If the answer is negative in step E122, then this step is followed by step E123 which is an initialization at which a first quantization symbol is considered in the current quantized vector SQV;

  The next step E124 is the entropy coding of the current symbol. The entropic coder that is used is that determined in step E116 for the group G to which the current symbol belongs. The group G, and its associated entropy coder are determined by the current value n.

  The next step E125 is a test to determine if the current quantization symbol is the last of the quantized current vector SQV; If the answer is negative, then this step is followed by step E126 at which the next quantization symbol is considered. The variable n is incremented by one unit to consider the next group. Step E126 is followed by step E124 previously described.

  When the response is positive in step E125 this step is followed by step E127 which is a test to determine if the current block is the last block to be processed. If the answer is negative, then this step is followed by step E128 to which a next block is considered. Step E128 is followed by step E119 previously described.

  When the response is positive in step E127, this step is followed by step E129 in which the distribution of the signal M representing the set of values of the variable M associated with the blocks is calculated. It is recalled that the variable M represents the number of quantization symbols respectively contained in each of the blocks.

  The following step E130 is the construction of an entropic coder for the signal M from its previously determined distribution. In a preferred embodiment, the entropy encoder is an arithmetic encoder. Alternatively, a Huffman encoder is used. Information for constructing the associated decoder is also written to the compressed file.

  The next step E131 is the coding of the signal M by the previously constructed entropy coder. The result is a bit stream that is written to the compressed file containing the encoding data of the image.

  The decoding of the coded image according to this embodiment is identical to the decoding previously described.

  Of course, the present invention is not limited to the embodiments described and shown, but encompasses, on the contrary, any variant within the scope of the skilled person.

Claims (2)

18 2862450 Claims   A method of encoding a digital signal representative of physical quantities comprising the steps of: dividing (E1) the signal into data blocks, transforming (E3) into frequency subbands of the data blocks, - quantizing (E4) ) transformed data blocks, for forming quantization symbol blocks, characterized in that it comprises the steps of: - forming (E8) groups of quantization symbols according to the rank of each symbol in its respective block, - construction (E18) of an entropy coding mode for each group of quantization symbols, - coding (E19) of each group by its respective entropy coder.   2. Method according to claim 1, characterized in that the transformation (E 3) is a DCT transformation of the data blocks.   3. Method according to claim 1 or 2, characterized in that the construction of an entropy coding mode for each group comprises for a given group: calculation (E17) of the distribution of the group, construction (E18) of a Entropic coding mode for the group based on its distribution.   4. Method according to any one of claims 1 to 3, characterized in that the entropy coding is an arithmetic coding.   5. Method according to any one of claims 1 to 4, the data blocks being two-dimensional, characterized in that it comprises the preceding step of transforming (E3) the data blocks into one-dimensional vectors.   6. Method according to any one of claims 1 to 5, characterized in that it comprises the steps of: - first division (E101) of the signal into data blocks, - first transformation (E103) into frequency sub-bands. data blocks, - first quantization (E104) of the transformed data blocks, 10 to form quantization symbol blocks, - construction (E116) of an entropy encoding mode for each quantization symbol group, - second division (E117) of the data block signal, - second transformation (E119) into frequency sub-bands of the data blocks, - second quantization (E120) of the transformed data blocks, to form quantization symbol blocks, - coding (E124) of each group by its respective entropy coder.   7. Method according to any one of claims 1 to 6, characterized in that the digital signal is a digital image.   8. A method of decoding a digital signal encoded by the method according to any one of claims 1 to 7, characterized in that it comprises the steps of: - construction of an entropy decoding mode respectively associated with each groups, - entropy decoding of the symbols of each of the groups, - formation of symbol blocks, according to the respective group of each symbol.   9. Device for encoding a digital signal representative of physical quantities comprising: 2862450 - means (20) for dividing the signal into data blocks; means (21) for transforming the data blocks into frequency sub-bands; means (22) for quantizing the transformed data blocks 5 to form quantization symbol blocks, characterized in that it comprises: means (23) for forming groups of quantization symbols according to the rank each symbol in its respective block; means (24) for constructing an entropy coding mode for each group of quantization symbols; means (25) for encoding each group by its respective entropy coder.   10. Device according to claim 9, characterized in that the means (21) of transformation are adapted to implement a DCT transformation of the data blocks.   11. Device according to claim 9 or 10, characterized in that the means (24) for constructing an entropy encoding mode for each group comprises: means for calculating the distribution of a given group; means for constructing an entropy coding mode for the group according to its distribution.   12. Device according to any one of claims 9 to 11, characterized in that the means (25) for entropy coding are adapted to implement an arithmetic coding.   13. Device according to any one of claims 9 to 12, the data blocks being two-dimensional, characterized in that it comprises means for converting the data blocks into one-dimensional vectors.   Apparatus according to any one of claims 9 to 13, characterized in that it comprises: - means of first division of the signal into data blocks, - means of first transformation into frequency sub-bands of data blocks, - first quantization means of the transformed data blocks, for forming quantization symbol blocks, - means for constructing an entropy encoding mode for each quantization symbol group, - second means dividing the signal into data blocks, - means of second transformation into frequency sub-bands of the data blocks, means of second quantization of the transformed data blocks, to form blocks of quantization symbols, - means of coding of each group by its respective entropy coder.   15. Device according to any one of claims 9 to 14, characterized in that it is adapted to process a digital signal which is a digital image.   16. Device for decoding a digital signal encoded by the device according to any one of claims 9 to 15, characterized in that it comprises: - means (40) for constructing a respective associated entropy decoding mode to each of the groups; means (41) for entropy decoding the symbols of each of the groups; means (42) for forming symbol blocks, as a function of the respective group of each symbol.   Coding device according to any one of claims 9 to 15, characterized in that the forming, construction and coding means are incorporated in: a microprocessor (100), a read-only memory (102) comprising a program for processing the data, and - a random access memory (103) having registers adapted to record modified variables during the execution of said program.   18. Decoding device according to claim 16, characterized in that the means of construction, decoding and formation are incorporated in: a microprocessor (100), a read-only memory (102) comprising a program for processing the data, and a random access memory (103) comprising registers adapted to record modified variables during the execution of said program.   19. Apparatus for processing (10) a digital image, characterized in that it comprises means adapted to implement the method according to any one of claims 1 to 8.   20. Apparatus for processing (10) a digital image, characterized in that it comprises the device according to any of claims 9 to 18.