FR2827408A1 - Genetic algorithm digital signal coding scheme - Google Patents

Abstract

A genetic algorithm digital signal coding scheme assigns coding parameters to sub signals in evolutionary steps comprising selection (E3) and reproduction (E4, E5, E6) from a set (E1) of mutated individuals with replacement by multipoint recombination (E5) using a number of points equal to the number of subsignals.

Description

<Desc / Clms Page number 1>

 The present invention relates to determining the value of the coding parameters of a digital signal divided into sub-signals.

 In the following, we will consider more particularly the case where the signal is an image to be coded according to the JPEG2000 standard, the description of which is available via the Internet at the address http: // www. jpeg. org. According to this standard, an image to be coded is divided into independent tiles, each tile having its own coding parameters.

 On the other hand, a known optimization technique is the evolutionary calculus described by Thomas Bâck, Ulrich Hammel and Hans-Paul Schwefel in Evolutionary Computation: Comments on the history and current state, IEEE Transactions on Evolutionary Computation, Vol. 1, N'l, April 1997.

 Evolutionary calculus (in English: evolutionary computation) is a research technique applied to problems which, on the one hand cannot be formulated mathematically continuously and on the other hand do not have a sufficiently reduced set of possible solutions to be able to be listed.

 Evolutionary calculus is thus part of the random techniques, and more specifically is part of the so-called guided random search techniques, that is to say which have a global or local function to optimize.

 If, in an analytical approach, we seek to design a single solution to solve a problem, evolutionary calculus considers

<Desc / Clms Page number 2>

 a set of solutions and evolves this set of solutions according to rules which are modeled on the evolution of life: reproduction, mutation and selection.

 The main idea is that multiple individuals can explore different strategies and combine their ability to converge on an effective solution. Reproduction allows combination, mutation allows exploration and selection allows convergence.

 Evolutionary calculus uses the following functions: - generation, which produces an initial population of individuals, - evaluation, which measures the performance of each individual with regard to the problem to be solved, - selection, which selects a subset of individuals among the current population, - reproduction, which produces a new set of individuals. Reproduction uses the combination of two selected individuals and the resulting mutation of the new individual. The mutation is the introduction of a slight difference in the individual child resulting from the combination, and - replacement, which replaces the individuals whose performance is the weakest with respect to the problem to be solved, by individual child generated by reproduction.

 These functions are used iteratively until a convergence criterion is satisfied.

 Examples of combinations are exposed by J. Culberson in Genetic invariance, a new paradigm for genetic algorithm design, technical report TR 92-02, June 1992, University of Alberta.

 This document describes monopoint recombination, multipoint recombination and uniform recombination (in English: uniform crossover).

 FIG. 9 illustrates the uniform recombination carried out on two vectors of parameters V1 and V2. Meeting points, here three in number, are distributed randomly over the vectors. A new vector is created by uniform recombination, by selecting the parameters of one of the vectors, here Vi, up to the first meeting point, then the parameters of the other vector, here V2, until the next meeting point , And so on.

<Desc / Clms Page number 3>

 The present invention aims to increase the speed of calculation of the coding parameters of a signal divided into sub-signals by providing a method and a device for determining the coding parameters of a digital signal.

 To this end, the invention proposes a method for determining the coding parameters of a digital signal representative of physical quantities, said signal being divided into sub-signals, a predetermined number of coding parameters being assigned to each of the sub-signals. , said determination being carried out by evolutionary calculation comprising the steps of: - generation of a set of individuals, each individual comprising a series of parameters for each of the sub-signals, each series comprising the predetermined number of coding parameters for the a respective of the undersignals, - evaluation of the individuals according to an evaluation criterion, - selection of a subset of individuals according to the result of the evaluation, - reproduction of the selected individuals, - replacement of the individuals not selected by individuals resulting from reproduction, characterized in that the reproduction step involves multi-recombination point of two individuals, according to which a number of meeting points is equal to the number of sub-signals and each meeting point is chosen randomly from one of the respective series of parameters.

 Thanks to the invention, the speed of convergence is higher compared to a combinatorial exploration consisting in testing all the combinations of possible parameters or, for the same calculation time, the parameters obtained are better.

 The meeting points are distributed uniformly across all sets of parameters, allowing rapid convergence to an effective solution.

<Desc / Clms Page number 4>

 According to a preferred characteristic, reproduction also comprises a step of mutation of the individuals resulting from reproduction.

 According to another preferred characteristic, the evaluation is carried out as a function of a coding cost.

 According to a preferred characteristic, the stages of evaluation, selection, reproduction and replacement are repeated. Thus, several sets of individuals are generated.

 According to a preferred characteristic, the method comprises the selection of the best individual evaluated during iterations. Thus, the best individual within the meaning of the evaluation criterion can be used for the subsequent coding.

 The invention also relates to a coding method characterized in that at least part of the coding parameters are determined by the previously presented determination method.

 Correlatively, the invention relates to a device for determining the coding parameters of a digital signal representative of physical quantities, said signal being divided into sub-signals, a predetermined number of coding parameters being assigned to each of the sub-signals, said determination being carried out by evolutionary computing means comprising: - means for generating a set of individuals, each individual comprising a series of parameters for each of the sub-signals, each series comprising the predetermined number of coding parameters for l '' a respective sub-signal, - means of evaluating individuals according to an evaluation criterion, - means of selecting a subset of individuals according to the result of the evaluation, - means of reproduction of selected individuals, - means of replacing individuals not selected by individuals resulting from reproduction, charac terrified in that the means of reproduction are adapted to effect a multipoint recombination of two individuals, according to which a

<Desc / Clms Page number 5>

 number of meeting points is equal to the number of sub-signals and each meeting point is chosen randomly from one of the respective sets of parameters.

 The invention also relates to a coding device characterized in that it is suitable for determining at least part of the coding parameters by the device previously presented.

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

 The invention also relates to a digital device including the device according to the invention, or means for implementing the method according to the invention. This digital device is for example a digital camera, a digital camcorder or a scanner. The advantages of the digital apparatus are identical to those previously exposed.

 The invention can be implemented by a computer program.

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

 The characteristics and advantages of the present invention will appear more clearly on reading a preferred embodiment illustrated by the attached drawings, in which: - Figure 1 is an embodiment of a device implementing the invention, - FIG. 2 is an embodiment of a device implementing the invention, - FIGS. 3 and 4 represent an image to be processed according to the invention, - FIG. 5 represents an embodiment of the method according to the invention, FIG. 6 represents the coding parameters of an image, FIG. 7 illustrates a recombination step included in the present invention, FIG. 8 illustrates a mutation step included in the present invention.

<Desc / Clms Page number 6>

 - Figure 9 illustrates a uniform recombination according to the prior art.

 According to an embodiment chosen and shown in FIG. 1, a data processing device according to the invention is a data coding device 2 which has an input to which a source 1 of non-coded data is connected.

 The source 1 comprises for example a memory means, such as random access memory, hard disk, floppy disk, compact disc, for storing uncoded 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.

 It will be considered more particularly below that the data to be coded is a series of digital samples representing an image.

 The source 1 supplies a digital image signal to the input of the coding device 2. The image signal is a series of digital words, for example bytes. Each byte value represents an image pixel, here at 256 gray levels, or black and white image. The images can be multispectral images, for example color images having components in three frequency bands, of the red-green-blue or luminance and chrominance type. Each band is then treated in a manner analogous to the monospectral image. As a variant, for color images, the hierarchical segmentation calculated on the luminance will be applied to the chrominance components.

 The data coding device comprises a device for determining the coding parameters of a digital signal representative of physical quantities, said signal being divided into sub-signals, a predetermined number of coding parameters being assigned to each of the sub-signals, said determination being carried out by an evolutionary calculation circuit comprising: a circuit 21 for generating a set of individuals, each individual comprising a series of parameters for each of the sub-signals, each

<Desc / Clms Page number 7>

 series comprising the predetermined number of coding parameters for a respective one of the sub-signals, - a circuit 22 for evaluating individuals according to an evaluation criterion, - a circuit 23 for selecting a subset of individuals as a function of the result of the evaluation, - a circuit 24 for reproducing the selected individuals, - a circuit 25 for replacing the unselected individuals with individuals resulting from the reproduction.

 According to the invention, the reproduction circuit 24 is adapted to perform a multipoint recombination of two individuals, according to which a number of meeting points is equal to the number of sub-signals and each meeting point is chosen randomly from one respectively series of parameters.

 The device 2 further comprises an identification circuit 26 and a coding circuit 27.

 The operation of the device 2 will be detailed below using an algorithm.

 The device according to the invention is integrated into a device, which is for example a digital camera, a digital camcorder, a scanner, a printer, a photocopier, a fax machine, a database management system, or even a computer. .

 Means 3 users of coded data are connected at the output 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.

 The storage and / or transmission means are connected to a decoding device 4 which decodes the data it receives. The decoded data, here forming a decoded image, are supplied to means 5 users of decoded data. The means 5 comprises for example a screen for displaying the decoded image.

<Desc / Clms Page number 8>

 FIG. 2 represents a device implementing the invention in the form of a microcomputer 10 connected to different peripherals, for example a digital camera 107 (or a scanner, or any means of image acquisition or storage ) connected to a graphics card and providing information to be processed according to the invention.

 The device 10 includes a communication interface 112 connected to a network 113 capable of transmitting digital data to be processed or, conversely, of transmitting data processed by the device. The device 10 also includes a storage means 108 such as for example a hard disk. It also includes a disc drive 109. This disc 110 can be a floppy disk, a CD-ROM, or a DVD-ROM, for example. The disk 110 like the disk 108 can 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 read-only memory 102 (called ROM in the drawing). In a second variant, the program can be received to be stored in an identical manner to that described previously 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 view the data to be processed or to serve as an interface with the user who can thus configure certain processing modes, using the keyboard 114 or any other means (mouse for example) .

 The central unit 100 (called 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. During power-up, the processing programs stored in a non-volatile memory, for example the ROM 102, are transferred to the RAM RAM 103 which will then contain the executable code of the invention as well as

<Desc / Clms Page number 9>

 registers to store 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 in the device, possibly removable, stores a program implementing the method according to the invention.

 The communication bus 101 allows communication between the different 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 capable of communicating instructions to any element of the microcomputer 10 directly or through another element of the microcomputer 10.

 As shown in FIGS. 3 and 4, the digital signal is a digital image which is to be coded according to a coding mode in which the image is divided into sub-signals. The actual coding will not be described here. For example, we use the JPEG2000 coding mode (from the English Joint Picture Expert Group), in which the image is divided into adjacent tiles.

The image represented comprises 512 by 768 pixels and is divided into six adjacent tiles Ti to T6 of 256 by 256 pixels.

 A tile T, is coded independently of the others, with its own coding parameters. In the embodiment chosen, each tile Ti has four coding parameters ai, bj, ci, dj, which are for example a sub-band filter number used from a set of predetermined usable filters, the number of resolution levels of the decomposition into sub-frequency bands, the number of bits used to represent the coded data of the tile and a post-processing number used from a set of predetermined usable post-treatments.

 FIG. 5 represents an embodiment of coding including the method for determining the parameters for coding a digital signal,

<Desc / Clms Page number 10>

 according to the invention. This method is implemented in the coding device and includes steps E1 to E11.

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

 The method according to the invention implements an evolutionary calculation comprising steps of generation, evaluation, selection, reproduction and replacement.

tuile, de la manière suivante : (a1, bi, ci, di, a2, b2, C2, d2,..., a6, b6, C6, d6) comme représenté à la figure 6. Les paramètres a1, b1, C1 et d1 sont relatifs à la tuile T1, et ainsi de suite. We consider the coding parameters of an image ordered by

tile, as follows: (a1, bi, ci, di, a2, b2, C2, d2, ..., a6, b6, C6, d6) as shown in Figure 6. The parameters a1, b1, C1 and d1 relate to tile T1, and so on.

 One such vector of ordered parameters is an individual, also called a solution. We consider a population of solutions, comprising N individuals, where N is an integer for example equal to 100.

 Step E1 is an initialization of the values of the parameters of the N individuals. For example, the parameter values are chosen at random. As a variant, the values of the parameters are equal to the default values indicated by the JPEG2000 standard. In the latter case, all individuals are identical.

 The next step E2 is an evaluation of the N individuals, using an evaluation method.

 The evaluation method is chosen according to the desired result. For example, to optimize the bit rate / distortion compromise, the evaluation criterion may be the cost of R + coding; . D, where R represents the bit rate of the coded image, D represents the distortion compared to the original image and  is a parameter set by the user.

<Desc / Clms Page number 11>

 As a variant, the evaluation criterion is the memory size used during decoding. According to another variant, a combination, linear or not, of several elements is used to assess an individual. It is thus possible for example to determine the parameters which offer a desired compromise between compression efficiency and calculation time.

 The next step E3 is a selection of the M best individuals in the population considered, within the meaning of the evaluation criterion, where M is an integer between 1 and the number of individuals. In our example, M is equal to 50.

 If the evaluation criterion is the R + I coding cost. D, then the 50 individuals with the lowest coding cost are selected here.

 The next step E4 is the selection of two individuals from the previously selected individuals. These two individuals are selected randomly, or alternatively in the order fixed by the evaluation operator: the first with the second during the first passage through step E4, the second with the third during the second passage through this step , and so on until the fiftieth with the first during the fiftieth passage through this step. As will be seen below, the number of iterations of steps E4 to E6 is limited, to fifty in this embodiment.

 The next step E5 is a multipoint recombination. For this, meeting points are determined randomly in the two individuals selected in the previous step. More precisely, a meeting point is determined randomly in each series of parameters corresponding to a tile, as shown in FIG. 7. There are therefore as many meeting points as there are tiles.

A new individual is produced from the two selected individuals, as follows:
In this new individual, also called a child, for a given series of parameters corresponding to a tile, the parameters located before the meeting point are those of the first selected individual and the parameters located after the meeting point are those of the second selected individual. Each tile is considered independently of the others for the

<Desc / Clms Page number 12>

 recombination. In other words, the border between two series corresponding to two tiles can be considered as a meeting point.

 The next step E6 is a mutation of the individual child generated in the previous step.

 For example, a parameter of the individual child is chosen randomly and then is modified randomly. In FIG. 8, the parameter a3 is chosen and a random value is added to it. The parameter a3 is replaced by the parameter as = as + v.

 The random value v can be determined so as to decrease as iterations progress.

 The next step E7 is a test to determine whether 50 individuals have been produced by the iterations of steps E4 to E6. As long as the answer is negative, then this step is followed by step E4 previously described.

 When the answer is positive in step E7, then this step is followed by step E8 in which the 50 newly produced individuals are inserted into the population so as to replace the 50 individuals not selected in step E3.

 The next step E9 is a test to determine whether a stop criterion is satisfied.

 The stopping criterion can be the number of iterations of steps E2 to E9. In this case, the number of iterations is for example fixed at 60. As a variant, the stopping criterion can be an objective to be reached for the evaluation of the best individual. For example, the cost of coding the best individual must be less than a predetermined threshold.

 As long as the stop criterion is not satisfied, then step E9 is followed by step E2 previously described.

 When the stop criterion is satisfied, then step E9 is followed by step E10 in which the best individual evaluated during all the iterations is identified.

 The best individual has the image coding parameters that will then be used.

<Desc / Clms Page number 13>

 The next step E11 is the coding of the image with the parameters previously defined. According to the invention, at least part of the coding parameters are determined as described above.

 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 (21)

1. Method for determining the coding parameters of a digital signal representative of physical quantities, said signal being divided into sub-signals, a predetermined number of coding parameters being assigned to each of the sub-signals, said determination being carried out by calculation evolutionary comprising the steps of: - generation (E1) of a set of individuals, each individual comprising a series of parameters for each of the sub-signals, each series comprising the predetermined number of coding parameters for the respective one of the sub-signals -signals, - evaluation (E2) of individuals according to an evaluation criterion, - selection (E3) of a subset of individuals according to the result of the evaluation, - reproduction (E4, E5, E6) of the selected individuals, - replacement (E8) of the unselected individuals with individuals resulting from the reproduction, characterized in that the reproduction step comprises a multi-recombination ipoint (E5) of two individuals, according to which a number of meeting points is equal to the number of sub-signals and each meeting point is chosen randomly from a respective one of the series of parameters.  2. Method according to claim 1, characterized in that the reproduction step further comprises a step (E6) of mutation of the individuals resulting from the reproduction.  3. Method according to claim 1 or 2, characterized in that the evaluation (E2) is carried out as a function of a coding cost. <Desc / CRUD Page number 15>  4. Method according to any one of claims 1 to 3, characterized in that the steps of evaluation, selection, reproduction and replacement are repeated.  5. Method according to claim 4, characterized in that the number of iterations is predetermined.  6. Method according to claim 4, characterized in that the number of iterations depends on a stopping criterion.  7. Method according to any one of claims 4 to 6, characterized in that it comprises the selection of the best individual evaluated during iterations.  8. Coding method characterized in that at least part of the coding parameters are determined by the determination method according to any one of claims 1 to 7.  9. Coding method according to claim 8, characterized in that at least one parameter is chosen from a sub-band filter number used from a set of predetermined usable filters, a number of resolution levels of a sub-decomposition frequency bands, a number of bits used to represent the coded data of a sub signal and a postprocessing number used among a set of predetermined usable postprocessings.  10. Device for determining the coding parameters of a digital signal representative of physical quantities, said signal being divided into sub-signals, a predetermined number of coding parameters being assigned to each of the sub-signals, said determination being carried out by evolutionary computing means including: <Desc / Clms Page number 16>  - means for generating a set of individuals, each individual comprising a series of parameters for each of the sub-signals, each series comprising the predetermined number of coding parameters for a respective one of the sub-signals, - means evaluation of individuals according to an evaluation criterion, - means of selecting a subset of individuals according to the result of the evaluation, - means of reproduction of the selected individuals, - means replacement of individuals not selected by individuals resulting from reproduction, characterized in that the means of reproduction are adapted to effect a multipoint recombination of two individuals, according to which a number of meeting points is equal to the number of sub-signals and each meeting point is chosen randomly from a respective one of the series of parameters.  11. Device according to claim 10, characterized in that the reproduction means are further adapted to effect a mutation of individuals resulting from reproduction.  12. Device according to claim 10 or 11, characterized in that the evaluation means are suitable for carrying out the evaluation as a function of a coding cost.  13. Device according to any one of claims 10 to 12, characterized in that the means of evaluation, selection, reproduction and replacement are adapted to reiterate their operation.  14. Device according to claim 13, characterized in that the number of iterations is predetermined. <Desc / Clms Page number 17>  15. Device according to claim 13, characterized in that the number of iterations depends on a stopping criterion.  16. Device according to any one of claims 13 to 15, characterized in that it comprises means for selecting the best individual evaluated during iterations.  17. Coding device characterized in that it is adapted to determine at least part of the coding parameters by the determination device according to any one of claims 10 to 16.  18. Coding device according to claim 17, characterized in that it is adapted to choose at least one parameter from a sub-band filter number used from a set of predetermined usable filters, a number of resolution levels of a decomposition into frequency sub-bands, a number of bits used to represent the coded data of a sub-signal and a post-processing number used from a set of predetermined usable post-treatments.  19. Device according to any one of claims 10 to 17, characterized in that the generation, evaluation, selection, reproduction and replacement 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) comprising registers adapted to store variables modified during the execution of said program.  20. Apparatus for processing (10) a digital signal, characterized in that it comprises means suitable for implementing the method according to any one of claims 1 to 9. <Desc / Clms Page number 18>  21. Apparatus for processing (10) a digital image, characterized in that it comprises the device according to any one of claims 10 to 18.