• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A system and method of processing for reconstructing a block of pixels using bins using a unit, a bit memory (MB), a pixel memory (MP) ). Said processing and calculation unit (UTC) being configured to receive a request for reconstruction of the block of pixels, to reconstruct a pixel of the support, said univocal pixel, from a bin read in the memory of bins (MB) for which there is a unique relationship and reconstruct each of the other pixels of the support, said recombination pixels, from the combination of a bin read in the bin memory (MB), said combination bin, the generation of which required the applying the multi-pixel transform of the block, with one or more pixels previously reconstructed and read in the pixel memory (MP), and writing into the pixel memory (MP), each recombination pixel and calculated.
    公开号:FR3019338A1
    申请号:FR1452585
    申请日:2014-03-26
    公开日:2015-10-02
    发明作者:Sylvain David；Pierre Evenou；Jean-Pierre Monchanin；Didier Feron
    申请人:Sylvain David；Pierre Evenou；Jean-Pierre Monchanin；Didier Feron；
    IPC主号:
    专利说明:

    [0001] The present invention generally relates to data processing of data. Distributed data storage systems typically employ massive replication of the stored data in order to achieve a satisfactory level of reliability. The use of erase code, such as the Mojette transform, provides a similar level of reliability while reducing the amount of redundant data required to reconstruct a block of data in a file. However, the known solutions of erasure codes have the disadvantage of being CPU time consuming, which restricts their use to specific uses such as backup. The object of the present invention is to propose a new system and a new data processing method that consumes less computing time. In particular, the object of the invention is to propose a new system and a new method of data processing making it possible to limit the time required for the reconstruction of a data block. To this end, the subject of the invention is a data processing system configured to reconstitute a block, called a support, formed of basic computer data, called pixels, by means of other computer data, called bins, resulting of a mathematical transform applied to said pixels, a number, denoted G, of groups of bins, called projections, having been generated from the pixels of said block such that said support can be reconstructed with a given minimum number, denoted M , of projections, M being less than G, characterized in that the system also comprises: - a data processing and calculation unit; a first memory zone, called a bins memory, in which said generated bins are stored; a second memory zone, called a pixel memory, distinct from the memory of bins and in which the pixels of said block to be reconstructed are intended to be written, said processing and calculation unit being configured to: a) receive a reconstruction request the block of pixels; b) reconstruct a pixel of the support, said univocal pixel, from a bin read in the memory of bins for which there exists a unique relation between this bin, called unequivocal bin, and this unequivocal pixel, and write in the memory of pixels, said univocal pixel of the block thus calculated, c) optionally, reconstructing other univocal pixels; d) reconstructing each of the other pixels of the support, said recombination pixels, from the combination of: a bin read in the bins memory, said combination bin, the generation of which required the application of the transform to several pixels of the block, with one or more pixels previously reconstructed and read in the pixel memory, and writing in the pixel memory, each recombination pixel thus calculated. Such a design of the system according to the invention makes it possible to optimize the memory accesses by rewriting each medium in the form of a bin or a combination of a bin and support (s) already reconstructed for n having to execute write operations only on the support memory in order to write the reconstructed supports and not having to carry out a multiplicity of successive reads / writes of the bins in the bins memory. Such a limitation on the number of write accesses of the bins memory makes it possible to increase the performance of reconstruction of the media with a single-core or multi-cc processor unit.
    [0002] Executing the calculation of the support directly by means of the projections without intermediate stages of execution of projection calculation allows the processor of the unit to use only queries of reading of the memory of bins where the projections are localized, without having to request new writes to this memory, and to write the calculation resulting from the medium directly on the second memory zone. In other words, such a configuration makes it possible for the calculation of a support to perform only one operation and thus to optimize the memory accesses. The optimization of memory access made by such a processing system 15 makes the processing system particularly suitable for systems and uses requiring a high bandwidth. Advantageously, the bins memory and the pixel memory are made in the form of two distinct physical supports. In particular, the bin memory may be part of a first computer or server device and the pixel memory may be part of a second computer or server device. The processing and computing unit may be part of the first or second device connectable by a communication network to the other device. According to a particular aspect, the processing and calculation unit is configured to reconstruct said pixels without writing operation in the bins memory during the reconstruction of said support pixels. The rewriting of the supports to be reconstructed in the form of a unique bin or of a combination of a bin with one or more supports makes it possible to be exempt during the reconstruction of intermediate calculations involving the writing of bins on the memory of bins for their update. Preferably, said combination operation corresponds to the application of the mathematical operator "exclusive OR" (XOR).
    [0003] The use of this mathematical operator makes it possible to limit the amount of memory necessary to execute the reconstruction operations of the support. According to an advantageous characteristic of the invention, the processing and calculation unit comprises a program, called a pixel reconstruction program, which makes it possible to carry out step d), said program presenting instructions according to which each pixel to be reconstructed, denoted support (k), other than that or those having an unequivocal relation to a bin, is calculated by an equation of the form: support (k) = projection [i] .bins [j] Asupport [I] or support (k ) = projection [i] .bins [j] Asupport [] ^ ... A support (n) with: k, I, n being indices defining different pixels of the support to be reconstructed; A: the XOR operator projection [i] .bins [j]: the value of a bin of the projection projection [i] that has been generated using the support pixel (k). In particular, the processing and calculation unit comprises another program, called a pixel reconstruction simulation program, including instructions: identification of a unique relationship between a pixel of the medium to be reconstructed and a bin of a projection ; - reconstruction of the corresponding pixel in a memory area of the unit 30 distinct from the memory of bins; runs through all the bins of at least M projections among the G projections generated to subtract, by application of the XOR operator, the value corresponding to this pixel to each bin generated using said pixel; repetition of the preceding steps for all the pixels of the support; and said processing and computing unit is configured to identify, during the execution of the simulation program, relations between the pixels and the bins, and generate said pixel reconstruction program based on the identified relationships between the pixels and the bins to define each pixel to be reconstructed, denoted support (k), other than that or those having an univocal relation to a bin, by an equation of the form: support (k) = projection [i] .bins [j] Asupport [I ] or support (k) = projection [i] .bins [j] Asupport [] ^ ... A support (n) Thus the system according to the invention makes it possible to generate previously on receipt of a reconstruction request, a code or program allowing, when executed after receiving a request, to reconstruct each medium without going through steps of intermediate writing of bins on the memory of bins. According to an advantageous characteristic of the invention, the unit is configured to make it possible to generate the G projections in the following manner: i) - organization of the block of pixels into a rectangle of dimensions {P, Q}, where P is the number of columns and Q the number of lines, then ii) - definition of a Cartesian coordinate system (p, q) whose origin is attached to a pixel preferably forming a corner of the rectangle, iii) - definition of G directions, said directions of projection, in the reference (p, q) iv) - for each of said G projection directions, for each pixel of a row or of a column of the block, the corresponding bin is calculated by combining said pixel with the other pixels of the block encountered in said projection direction or in equality with said pixel when no other pixel is encountered in said projection direction. In particular, the XOR operator is applied to the pixels of the block encountered in the direction of projection. Advantageously, concerning the reference (p, q), the vector p corresponds to the axis of the columns and the vector q to that of the lines.
    [0004] The number G of projections generated corresponds to the sum of the minimum number M of projections necessary to reconstruct the generated support and the number, denoted n, corresponding to the desired level of redundancy.
    [0005] The minimum number M of projections and the number G of projections to be generated are defined by the user at the time of application of the transforms to the pixels of the support. According to an advantageous characteristic of the invention, to generate a bin corresponding to a given combination of pixels, the processing and calculation unit is configured to determine, for the bin to be generated, the necessary combinations of pixels; and the processing and computing unit is configured to generate a program, referred to as a bin generation program, which comprises the following instructions: - reading pixels to be combined with one another from the pixel memory; generating a bin by combining the values of said pixels with each other; writing the result of the previous operation in the bins memory, or, when there exists a unique relation between a pixel and a bin to be generated: reading of said pixel; generation of the bin from said pixel and writing of said bin in the bins memory, and the program is free of instructions for resetting the bins values to be generated, of reading step, said intermediate reading, of the bin; generating and combining step between a bin to be generated and said pixels. With such a design of the system according to the invention, it is no longer necessary to perform a number of successive reads and writes as important as with the known solution of the state of the art to generate a bin.
    [0006] According to an advantageous characteristic of the invention, the processing and calculation unit comprises another program, called the bins generation simulation program, including instructions for resetting the bins values to be generated, read, read intermediate, of the bin to be generated and of combination between a bin to be generated and said pixels, and the processing and calculation unit (UTC) is configured to generate said bins generation program after having executed the simulation generation program of bins, the unit being configured to generate said bins generation program by identifying the pixel (s) used by said bins generation simulation program to generate the corresponding bins and defining pixel combination instructions which are free of said instructions resetting the bin values to be generated, the reading step, said intermediate reading, the bin to generate and combining step between the bin to generate and said pixels.
    [0007] In other words, said bins generation simulation program, which corresponds to the known program of the state of the art for generating a bin, is carried out "off-line", that is to say without writing on the support memory. and before receiving a bin generation request, to generate a new bin generation program, free from memory access consuming instructions for it to be executed "online", i.e. in connection with the support memory and following receipt of a request for generation of bins.
    [0008] In other words, the offline pre-execution of the inverse and direct Mojette transforms, known from the state of the art, enables the system according to the invention to generate new programs, intended to be executed online, which take into account the memory access required for the respective transforms. The invention also relates to a method of treatment and a corresponding computer program product. For this purpose, the invention proposes a data processing method configured to reconstitute a block, called a support, formed of basic computer data, called pixels, by means of other computer data, called bins, resulting from a mathematical transform applied to said pixels, a number, denoted G, of groups of bins, called projections, having been generated from the pixels of said block so that said support can be reconstructed with a given minimum number, denoted M, of projections , M being less than G, characterized in that the method comprises the following steps: a) reception by a data processing and calculation unit of a reconstruction request of the block of pixels; b) reconstruction by the processing and calculation unit, of a support pixel, said univocal pixel, from a bin read in a memory, called a bin memory, for which there is a unique relationship between this bin, called unequivocal bin, and this univocal pixel, and write in a memory, called pixel memory, distinct from the memory of bins, said univocal pixel of the block thus calculated, c) optionally, reconstruction of other univocal pixels; d) reconstruction using the processing unit and calculating each of the other 25 pixels of the support, said recombination pixels, from the combination of: a bin read in the memory of bins, said bin of combination, the generation of which required the application of the multi-pixel transform of the block, with one or more pixels previously reconstructed and read in the pixel memory, and writing in the pixel memory, of each recombination pixel thus calculated.
    [0009] According to a particular aspect, the reconstruction of said pixels is performed without writing operation in the bins memory during the reconstruction of said support pixels.
    [0010] According to an advantageous characteristic of the invention, said combination operation corresponds to the application of the mathematical operator "exclusive OR" (XOR).
    [0011] According to an advantageous characteristic of the invention, for the realization of step d), each pixel to be reconstructed, noted support (k), other than that or those having a unique relationship to a bin, is calculated by an equation of the form: support (k) = projection [i] .bins [j] Suppport [I] OR support (k) = projection [i] .bins [j] Suppport [I] ^ ... Support (n) with: k, I, n being indicia defining different pixels of the medium to be reconstructed; : the operator XOR projection [i] .bins [j]: the value of a bin of the projection projection [i] that has been generated using the support pixel (k). According to an advantageous characteristic of the invention, prior to receiving the reconstruction request, the reconstruction of the pixels is simulated at the level of the processing and calculation unit by executing steps comprising: identification of a unique relationship between a pixel of the support to be reconstructed and a bin of a projection; reconstruction of the corresponding pixel in a memory zone of the unit 30 distinct from the memory of bins; runs through all the bins of at least M projections among the G projections generated to subtract, by application of the XOR operator, the value corresponding to this pixel to each bin generated using said pixel; repetition of the preceding steps for all the pixels of the support; then, from the identified relationships between the pixels and the bins, each pixel to be reconstructed, noted support (k), other than that or those having a unique relationship to a bin, is defined in the processing unit and calculation by an equation of the form: support (k) = projection [i] .bins [j] Suppport [I] OR support (k) = projection [i] .bins [j] Suppport [] ^ ... A support (n According to an advantageous characteristic of the invention, the G projections are generated in the following manner: i) - organization of the block of pixels into a rectangle of dimensions {P, Q}, where P is the number of columns and Q is the number of lines, then ii) - definition of a Cartesian coordinate system (p, q) whose origin is attached to a pixel preferably forming a corner of the rectangle, iii) - definition of G directions, referred to as projection directions, in the reference frame (p, q) iv) - for each of said G projection directions, the corresponding bin is calculated for each pixel of a row or column of the block waving by combining said pixel with the other pixels of the block encountered along said projection direction or by equality with said pixel when no other pixel is encountered along said projection direction. According to an advantageous characteristic of the invention, the number G of projections generated corresponds to the sum of the minimum number M of projections required to reconstruct the generated support and the number, denoted n, corresponding to the desired level of redundancy. According to an advantageous characteristic of the invention, to generate a bin corresponding to a given combination of pixels, the necessary combinations of pixels are determined for the bin to be generated; and the following steps are performed: - reading pixels to be combined with one another from the pixel memory; generating a bin by combining the values of said pixels with each other; writing the result of the preceding operation in the bins memory, or, when there is a one-to-one relation between a pixel and a bin to be generated, reading said pixel; generation of the bin from said pixel and writing of said bin in the bins memory, and said bin is generated without resetting steps of the bin values to be generated, reading step, said intermediate reading, of the bin to be generated and 10 of combining step between a bin to be generated and said pixels. According to an advantageous characteristic of the invention, prior to the generation of bins, a generation of bins is simulated which includes instructions for resetting the bins values to be generated, read, referred to as intermediate reading, of the bin to be generated and of combining a bin to generate and said pixels, and a bins generation program is generated after simulating the generation of bins, said bins generation program being generated by identifying the pixel (s) used in the generation simulation of bins 20 for generating the corresponding bins and defining pixel combining instructions which are free of said reset instructions of the bin values to be generated, read step, intermediate read, bin to be generated and step combining the bin to generate and said pixels. The invention also relates to a computer program product downloadable from a communication network and / or stored on a computer readable medium and / or executable by a processor of a processing and calculation unit, characterized in that it comprises program code instructions for carrying out the steps of a method as described above when said program is executed by said processor of the unit.
    [0012] The invention will be better understood on reading the following description of exemplary embodiments, with reference to the appended drawings in which: FIG. 1 is a view of an installation according to the invention which comprises a pixel memory, a memory of bins and a processing and calculation unit; FIG. 2 is a view illustrating the generation of bins from the pixels of a support along several directions of projections denoted by projection 0, projection 1, projection 2, the index of bins obtained being denoted bin_idx; FIG. 2A is a table showing the composition of each projection bin generated as illustrated in FIG. 2; FIG. 3 illustrates the various read-write and write exchanges of data between the unit and the memories for calculating a bin, denoted bin [0], according to the transform direct Mojette known from the state of the art; FIG. 4 details the steps of execution of the direct Mojette transform according to the invention; FIG. 5 illustrates the reconstruction path for each of the possible projection combinations with the projections projection 0, projection 1 and projection 2 obtained corresponding to FIG. 2; FIG. 6 illustrates the order corresponding to the arrows, in which the bins of the combination of projection 0 and projection 1 projections are traveled to reconstruct two pixels; FIG. 7 illustrates the exchanges between the unit (cpu) and the memories for the reconstruction of two pixels (pixel [0] and pixel [128]) of the support from the projection bins 25 and the projection bins the inverse Mojette transform known from the state of the art; FIG. 8 illustrates the operations carried out for the same example as that of FIGS. 6 and 7 by applying the inverse Mojette transform according to the invention. In the example detailed below and illustrated in the figures, a mathematical transform is applied to a block of computer data. The basic computer data of said block are called pixels. In other words, the pixels form the basic information of the block. The block is called support. The mathematical transform applied to the pixels generates several groups of computer data, called bins. Said groups are called projections. The bins of a first group are distinguished from the bins of a second group by the direction of projection used to generate the corresponding bins. A projection is a set of bins to which a given projection direction corresponds. Thus, a projection is defined by a size in bins, and an angle (p; q) defining a projection direction as detailed below. The transform according to the invention applied in the sense of a reconstruction of a support from the bins of several projections or a generation of bins, in particular of several projections, from the pixels of a support is based on on the Mojette transform which is detailed below. The installation and the method according to the invention use a modified Mojette transform which will be detailed later. In general, for any further information on the Mojette transform and its properties, the reader will usefully refer to the article "Internet distributed image information system" by JP Guédon, B. Parrein and N. Normand, published in "Integrated Computer - Aided Engineering", 8 (2001), 25 pages 205-214, ISSN 1069-2509, las Press, which is fully incorporated by reference in the description. Subsequently, the following appellations will be used: -transformed current direct Mojette: the known transform of the state of the art applied to a block of pixels to decompose it into bins, -the direct Mojette transform according to the invention: the transformed according to the invention, corresponding to a modification of the current Mojette transform, applied to a block of pixels to decompose it into bins, - current inverse Mojette transform: the known transform of the state of the art applied to the bins for reconstructing the block of pixels, - direct Mojette transform according to the invention: the transform according to the invention, corresponding to a modification of the current inverse Mojette transform, applied to the bins to reconstruct the block of pixels.
    [0013] Concerning more particularly the implementation of the current inverse Mojette transform, the reader will usefully refer to the article "A geometry driven reconstruction algorithm for theMojette transform", from Nicolas Normand, Andrew Kingston, Pierre Evenou, IRCCyN / IVC UMR CNRS 6597 "published in" Discrete Geometry for Computer Imagery, Szeged: Hungary (2006) which is fully incorporated by reference in the description The reader may also refer to Appendices A and B of B.Parrein's thesis "Multiple description of the information by transformation Mojette, Benoît Parrein, doctoral thesis, University of Nantes, November, 2001 "which can be consulted in particular at the following address: http://www.irccyn.ec-nantes.fr/--parrein/pubs. htm and which is fully incorporated by reference in the description.For more particularly the implementation of the current direct Mojette transform, the reader will usefully refer to chapter 3 3.2.1 25 of B.Parrein's thesis cited above. The block of pixels is in the form of a queue of pixels that can be considered for its processing as a rectangle of dimensions {P, Q}. P is the number of columns and Q is the number of rows. Advantageously, the number of lines Q corresponds to the minimum number M of projections necessary for the reconstruction of the block of pixels.
    [0014] A Cartesian coordinate system (p, q) is associated with this rectangle. The origin of this marker is attached to a pixel of the rectangle, preferably a pixel located at a corner of the rectangle. The vector of this marker corresponds to the axis of the lines. The vector q corresponds to the axis of the columns.
    [0015] Projection directions are defined by composition of the two integers (p, q). For example, the projection direction p = 0 and q = 1 corresponds to a projection in the direction of the columns.
    [0016] A projection operation consists, for a projection direction defined by a pair of values p, q, to scan a line or a column and, for each pixel of the scanned line or conne, to combine said pixel with the other pixels encountered along this direction of projection.
    [0017] In practice, a displacement in the block of pixels in the direction of projection corresponds to a displacement in the queue of pixels with a jump value, called offset, given. In the example illustrated in the figures, the combination of two elements between them, for example two pixels, consists of the application of a mathematical operator, preferably the operator or exclusive, that is to say the XOR operator. The number, denoted M, of necessary minimum projection directions desired for the reconstruction and the desired level of redundancy n are initially defined by the user when he initiates the projection generation method (i.e. groups of bins) using the system as described below. A metadata file is associated with the projections generated to, during the reconstruction, be able to determine the numbers M and n used 30 and possibly the size of the block. The level of redundancy desired corresponds to the number of additional projections that it is desired to generate in addition to the number of projections (greater than or equal to 2) defined as being necessary to reconstruct the block of pixels. For a desired redundancy level equal to n, the minimum number of projections to be applied to the pixel block to be able to reconstruct it with said desired redundancy level is M + n. Thus, the current direct Mojette transform generates M + n projections. In such a current implementation, the generation of a bin, denoted here "proj (k)", of a projection, is written in the general case and for a projection 10 of angles (p, q), with a equation of the type: proj (k) = support (i, j) xor support (I, r) .... k being an integer, with proj (k) being the kth bin of the projection and support (i, j) and support (I, r) being pixels of the support, i, j and I, r being indices defining the position of the pixel in the rectangle of pixels of dimensions {P, Q}. The terms support (i, j) and support (I, r) can also be written pixel [x] and pixel [y], x and y corresponding to the rank of these pixels. Similarly, if the projection is denoted 20 projection0, proj (k) can be written projection0.bin [k]. In particular, the current implementation of the Mojette Transform is written: start For each line from 1 to Q make 25 For each column from 1 to P make k <- p.line - q.column proj (k) xor <- support (line, column) endTo endFor 30 End As recalled above, a projection "proj" thus corresponds to a set of bins "(proj (k)". In particular said projection "proj" is defined by a size in bin, an angle (p; q) The number of pixels to be taken into account for the generation of a bin depends on the angle (p; q) associated with the projection.
    [0018] However, as detailed in the example below, to generate such a projection of bins proj (k), the implementation of the current direct Mojette transform, that is to say known from the state of the art , plans to pre-initialize all bits of a projection to 0 before proceeding to the generation of a direct current Mojette transform projection.
    [0019] In the detailed example below, the current direct Mojette transformation is applied on a 256-pixel support. In this example we consider the case where M = 2 and n = 1. Three projections are thus generated. The pixel support is divided into two lines (M = 2) of 128 pixels each. Thus we have a rectangle {P, Q} of 128 columns and 2 lines. The angles (p; q) associated respectively with the projections to be generated are the following: projection: p = 1; q = -1; - projection1: p = 0; q = 1; - projection2: p = 1; q = 1; Figure 2 illustrates the application of the direct Mojette transform on a pixel rectangle (2,128) and the representation of the three resulting projections, as well as the associated bins values. The table of FIG. 2A details the contents of each of the projections generated in FIG. 2 by the current direct Mojette transform on the 256 pixel support.
    [0020] As illustrated in FIG. 1, the current direct Mojette transform, as well as the direct Mojette transform according to the invention detailed hereinafter, can be implemented using an installation which comprises: a first memory, called MP pixel memory: this MP 5 memory contains the pixels of the medium on which the direct Mojette transform operation is to be applied; a UTC processing and calculation unit: the unit comprises a processor, referenced cpu or PROC in certain figures, which performs the read / write operations to the pixel and bit memories, as well as application operations. of a mathematical operator, here the XOR operator, between the pixels. As detailed below, the processor can also apply this operator to a combination of bin (s) and pixel (s). a second memory, called memory of bins MB: this memory zone is used to store the results (bins) of the direct Mojette transform. The processing and calculation unit UTC comprises one or more working memories associated with the processor. The unit can be implemented as a computer. The memories and the unit of the computer installation described above can also be used to execute the direct Mojette transform according to the invention, the current inverse Mojette transform and the inverse Mojette transform according to the invention. The contents of the bins memory may be extracted by another entity. This entity can store said bins, for example, on a disk separate from the binary memory or broadcast them over a network, for example in the form of UDP / IP frames. Direct Direct Mojette Transformation Execution steps of the current direct Mojette transform are detailed below in the context of calculating and writing a bin corresponding to the index 0, namely the bin [0 ], bins of projection1 shown in Figure 2.
    [0021] Thus, in connection with FIG. 3, which illustrates the various exchanges between the processor and the memories for calculating bin [0] (or more precisely projection1.bin [0]), said calculation of bin [0] comprises the following steps: 1) resetting the bin [0] to be generated; 2) reading the pixel [0] from the pixel memory; 3) reading the bin [0] of the projection1; 4) XOR between the current contents of the bin [0] of the projection1 and the value of the pixel [0]; 5) writing the intermediate result of the previous operation in the memory of bins; 6) reading the pixel [128] from the pixel memory; 7) replay of the bin [0] of the projection1 from the memory of bins; 8) XOR between the current content of the bin [0] of the projection1 and the value of the pixel [128]; 9) Write the final of the previous operation in the memory of bins. In the context of the example chosen, to obtain the complete transformation of the support, this set of operations is repeated 128 times.
    [0022] Concerning the generation of the two other projections, namely the "projection" and "projection" projections, the process is performed in a similar way by combining the pixels encountered in the projection direction applied for the "projection" or the "projection". . The choice of pixels to be combined depends on the angle associated with the projection (see Figure 2).
    [0023] Direct Mojette Transformation According to the Invention The implementation of the direct Mojette transform according to the invention makes it possible to exempt the steps of successive reads / writes in the generation of a bin of a projection, and to suppress the initialization of the content. of a projection.
    [0024] Execution steps of the direct Mojette transform according to the invention are detailed below with reference to FIG. 4 in the context of calculating and writing a bin corresponding to the index 0 (bin [ 0]) bins of the projection1. Thus, computing the bin [0] (or more precisely projection1.bin [0]) by applying the direct Mojette transform according to the invention comprises the following steps: 1) reading the pixel [0] from the memory pixels; 2) reading the pixel [128] from the pixel memory; 3) XOR between the current content of the bin [0] of the projection 1 and the value of the pixel [128]; 4) Writing the result of the previous operation in the pixel memory For the implementation of the Direct Mojette Transform according to the invention which makes it possible to avoid intermediate read / write operations, the execution unit comprises a code generator. The code generator is configured to perform the following operations: 1) executing the current direct Mojette transform algorithm to determine for each bin of a projection the necessary pixel combinations of said bin; 2) production of the instructions necessary for the generation of a bin in the form: proj (k) = support (i, j) xor support (I, p) .... 30 Where support (i, j), support ( I, p) are the pixels involved in the generation of the proj (k) bin.
    [0025] The code generator can generate as many direct Mojette transform programs according to the invention as there are user block size and torque combinations (M, n) configured.
    [0026] When the processing and calculation unit receives a request to decompose the block of pixels into bins, the unit executes the corresponding program generated by the code generator in order to apply the Mojette transform according to the invention to the block. pixels.
    [0027] After execution of said program, a set of 364 bins is obtained corresponding to the bins of the projections "projection 0", "projectionl" and "projection 2" as illustrated in FIG. 3A. The set of bins thus obtained is the same as with the current direct Mojette transform, but in a much faster way since the number of operations necessary to generate these bins with the program (s) produced by the generator of code, is weaker. To reconstruct the support, the unit performs a reverse Mojette transformation detailed below.
    [0028] Current Reverse Mojette Transformation The current inverse Mojette transformation, i.e., known from the state of the art, is described below to then show the modifications made to this current inverse Mojette transformation in order to obtain inverse Mojette transformation according to the invention. We start from a univocal correspondence between a pixel of the support and a bin of a projection. The corresponding univocal pixel is reconstructed and the processor traverses all the projections associated with the medium to subtract the value corresponding to that pixel to each bin generated using said pixel. This process is repeated for all the pixels of the support.
    [0029] The subtraction operation is performed by applying the XOR operator between the pixel whose value is to be subtracted and the bin. Indeed, assuming that the contents of a memory cell (bin [k]) of the bin memory contain the result of an XOR operation between the pixel pixel [i] and the pixel pixel [j], and Knowing the pixel value [i], we obtain the pixel value [j] by the following operation: pixel [j] = pixel [i] xor bin [k] 10 To illustrate the reconstruction process, we start from example, previously described, so that the medium to be reconstructed is the support of 256 pixels from which 3 projections were generated. Thus, the bit memory MB contains the bins of the three previously generated projections and the processing and calculation unit UTC is used to reconstruct said medium. The metadata file associated with the generated projections makes it possible to determine that only two of the three projections generated to reconstitute the support are sufficient (for recall M = 2 and n = 1). Figure 5 illustrates the reconstruction path for each of the combinations. possible projections. 25 -combination {projection 1, projection 0): projection0.bin [0] contains the index pixel 128; -combination {projection 2, projection 0): projection0.bin [0] contains the index pixel 128 and the projection projection1.bin [0] contains the pixel of index O. -combination {projection 2, projection 1) : projection2.bin [0] contains the index pixel 0 To start the reconstruction, the processing and calculation unit starts from a bin for which there is an unambiguous correspondence between the bin and a pixel of the support.
    [0030] An example of reconstruction of the initial support from projection1 and projection ° is presented below. As shown in FIG. 5, the starting point is the bin of index 0 of the projection 0 (projection0.bin [0]) which contains the index pixel 128.
    [0031] The index pixel 0 is obtained by an operation XOR between the index pixel 128 and the index bin 0 of the projection 1 (projection1.bin [0]): pixel [0] = pixel [128] XOR projection1 bin [0].
    [0032] This process is iteratively repeated just at index bin 127. The operations performed to reconstruct the support pixels from bin of index i are: - pixel [i + 128] = pixel [i-1] xor projO.bin [i] - pixel [i] = pixel [i + 128] xor projl.bin [i] As in the case of the direct Mojette transform described above, the implementation of the inverse Mojette transform current is carried out with a memory of bins, a memory of pixels and a unit of an installation as presented above.
    [0033] To begin the operation of reconstructing the pixels of the support, the memory of bins MB is loaded previously with the necessary number of projections to carry out a reconstruction. In the example used in the document (M = 2, n = 1), this bins memory is loaded with 2 projections, that is to say with 256 bins. The source of these projections may be, for example, a magnetic medium, a disk or a frame of an IP network. In the case of the current inverse Mojette transform, the unit is configured as detailed hereinafter to perform the operations necessary to reconstruct the support pixels from the projections by performing reads / writes to the bin memories and pixels. FIG. 7 illustrates the exchanges between the unit and the memories for the reconstruction of two pixels (pixel [0] and pixel [128]) of the support from projection bins 10 and projection bins 10. The operation described below corresponds to the operation. Figure 6 illustrates the order corresponding to the arrows in which the bins are traveled to reconstruct the two pixels. In connection with FIG. 7, the details of the operations performed by the current implementation to reconstruct the two pixels is as follows: 1) reading of the content of the bin [0] of the projection ° which contains the pixel [128] (datum ), 2) writing the pixel [128] in the support memory; 3) set to zero the value of the bin just read (XOR operation on itself); 4) writing of the result in the memory support (bin of index 0 projection0); 5) reading the bin [0] of the projection 1 (which contains the XOR result between the pixel [0] and the pixel [128]); 6) XOR operation allowing the reconstruction of the pixel [0] (data 2); 7) writing of the result of the previous operation in the bin [0] of the projection 1: the written value corresponds to the value of the pixel [0]; 8) reading of the bin [0] of the projection 1: contains the value of the pixel [0] written in the previous operation; 9) writing of the previously read value in the support memory (pixel [0]); 10) set to zero the value of the bin just read (XOR operation on itself); 11) writing of the result in the memory support (bin of index 0 projection1); 5 12) reading the bin of index 1 of the projection °; 13) XOR operation between the content just read and the value of the pixel [0]: the result obtained is the value of the pixel [129]; 14) Writing the result of the previous XOR operation in the bin [1] of the projection ° 10 In the case of the implementation of the current Mojette transform described in the example above, the reconstruction of a pixel involves two XOR operations and two bin rewrites of a projection. In a general way we have M rewrites of bin of a projection. Inverse Mojette Transform According to the Invention In the inverse Mojette transform object of the invention, these intermediate operations (read / write in the memory of bins) are suppressed. This results in a faster reconstruction based on the fact that, except for the unambiguous pixel (s), the reconstruction of a pixel of the medium requires a combination between the bin of a projection and a number of pixels of already reconstructed support. This number depends on the number M of projections defined. Thus, compared to the previous example, pixel [128] pixel [0] and pixel [129] can be reconstructed as follows: pixel [128] = projections [0] .bins [0] pixel [0] = projections [1] .bins [0] Apixel [128] and pixel [129] = projections [0] .bins [1] Apixel [0]] Indeed, projections [0] .bins [0] contains the univocal pixel ( pixel [128]) and thus constitutes the starting point of the reconstruction. Thus, in the new implementation of the inverse Mojette transformation, the reconstruction of a pixel [k] of a medium is given by the general form: pixel [k] = projections [i] .bin [j] Asupport [ 1] ^ ... Suppport [n]; FIG. 8 illustrates the operations carried out for the same example by applying the new inverse Mojette transform, object of the invention: 1) reading the content of the bin [0] of the projection ° which contains the pixel [128] (datum 2) writing of the pixel [128] in the support memory 3) reading of the bin [0] of the projection1 (contains the result of the XOR between the pixel [0] and the pixel [128]; 4) reading of the pixel [128] from the support memory; 5) XOR between the pixel [128] and the content of the bin [0] of the projection1: the result gives the value of the pixel [0]; 6) writing of the result in the memory support (pixel [0]). In order to generate a reconstruction program using the Inverse Mojette transforms of the invention, the unit is configured to perform a pixel reconstruction program using the current inverse Mojette transformation, in "off-line" mode. that is to say by simulating the steps of writes on the memories of bins and pixels. This execution makes it possible to obtain for each pixel to reconstruct the pixels and the bin of the projection involved in its reconstruction or, in the case of a univocal relationship, to obtain a unique pixel corresponding to a unique bin. From the relationships between pixels and bins identified during said offline execution, the unit generates one or more reconstruction program (s) called "online" reconstruction program (s) according to which each pixel to be reconstructed, noted pixel [k], other than that or those having a unique relationship to a bin, is computed by an equation of the form: pixel [k] = projection [i] .bin [j] Apixel [1] or pixel [k ] = projection [i] .bin [j] Apixel [I] ^ ... ^ pixel [n] with: k, I, n being indices defining different pixels of the medium to be reconstructed; : the operator XOR projection [i] .bin [j]: the value of a bin that has been generated using the pixel [k], in particular the value of the jth projection bin [i]. The unit can generate as many reconstruction programs as there are combinations of possible projections and possibly size of configured blocks. The installation and the method according to the invention can be used for various applications including "Scale-out NAS-Cloud storage", "Networking", ad-hoc network, image coding and video coding applications. Each memory may be formed of several distinct and decentralized memory zones or not. The present invention is in no way limited to the embodiments described and shown, but those skilled in the art will be able to make any variant within their mind. Those skilled in the art readily understand that the various steps and functions of the above embodiments can be embodied as computer programs. In particular, the steps described above can be performed in the form of electronic and / or computer instructions executable by a processor.
    [0034] These computer programs, or computer instructions, may be contained in program storage devices, such as computer-readable digital data storage media, or executable programs. Programs or instructions can also be run from program storage devices. The present invention is in no way limited to the embodiments described and shown, but the person skilled in the art will be able to make any variant 10 in accordance with his spirit.
    权利要求:
    Claims (19)
    [0001]
    REVENDICATIONS1. A data processing system configured to reconstruct a block, called a medium, formed of basic computer data, called pixels, using other computer data, called bins, resulting from a mathematical transform applied to said pixels, a number , denoted G, of groups of bins, called projections, having been generated from the pixels of said block so that said support can be reconstructed with a given minimum number, denoted M, of projections, M being less than G, the system also comprising: - a data processing and calculation unit (UTC); a first memory zone, called a bit memory (MB), in which said generated bits are stored; is - a second memory zone, called pixel memory (MP), distinct from the bins memory (MB) and in which the pixels of said block to be reconstructed are intended to be written, characterized in that said processing and calculation unit (UTC) is configured to: a) receive a request to reconstruct the block of pixels; b) reconstruct a pixel of the support, said univocal pixel, from a bin read in the memory of bins (MB) for which there is a unique relationship between this bin, called univocal bin, and this univocal pixel, and write in the pixel memory (MP), said univocal pixel of the block thus calculated, c) optionally, reconstructing other univocal pixels; d) reconstructing each of the other pixels of the support, said recombination pixels, from the combination of: a bin read in the bin memory (MB), called bin of combination, the generation of which required the application of the transformed to several pixels of the block, with one or more pixels previously reconstructed and read in the pixel memory (MP), and writing in the pixel memory (MP), each recombination pixel thus calculated .
    [0002]
    2. System according to claim 1, characterized in that the processing unit 5 and calculation (UTC) is configured to reconstruct said pixels without writing operation in the memory of bins (MB) during the reconstruction of said pixels of the support.
    [0003]
    3. System according to one of the preceding claims, characterized in that said combination operation corresponds to the application of the mathematical operator "exclusive OR" (XOR).
    [0004]
    4. System according to one of the preceding claims, characterized in that the processing and calculation unit (UTC) comprises a program, referred to as a pixel reconstruction program, for carrying out step d), said program presenting instructions according to which each pixel to be reconstructed, denoted as medium (k), other than that or those having an unambiguous relation to a bin, is calculated by an equation of the form: support (k) = projection [i] .bins [ j] Asupport [] 20 or support (k) = projection [i] .bins [j] Asupport [] ^ ... A support (n) with: k, I, n being indices defining different pixels of the support to be reconstructed ; 25 ^: the operator XOR projection [i] .bins [j]: the value of a bin of the projection projection [i] that has been generated using the support pixel (k).
    [0005]
    5. System according to claim 4, characterized in that the processing unit 30 and calculation (UTC) comprises another program, called pixel reconstruction simulation program, including instructions: - identification of a unique relationship between a pixel of the support to be reconstructed and a bin of a projection; - reconstruction of the corresponding pixel in a memory area of the unit separate from the memory of bins; runs through all the bins of at least M projections among the G projections generated to subtract, by application of the XOR operator, the value corresponding to this pixel to each bin generated using said pixel; repetition of the preceding steps for all the pixels of the support; and in that said processing and computing unit is configured to identify, during execution of the simulation program, relations between the pixels and the bins, and generate said pixel reconstruction program based on the identified relationships between the pixels. pixels and bins to define each pixel to be reconstructed, noted support (k), other than that or those having an unequivocal relation to a bin, by an equation of the form: support (k) = projection [i] .bins [j ] Asupport [1] 15 OR support (k) = projection [i] .bins [j] Suppport [] ^ ... Support (n)
    [0006]
    6. System according to one of the preceding claims, characterized in that the unit is configured to allow generating the G projections in the following manner: i) - organization of the block of pixels into a rectangle of dimensions {P, Q }, Where P is the number of columns and Q is the number of lines, then ii) - definition of a Cartesian coordinate system (p, q) whose origin is attached to a pixel preferably forming a corner of the rectangle, iii) definition of G directions, referred to as the projection directions, in the reference (p, q) iv) for each of said G projection directions, for each pixel of a row or a column of the block, the bin is calculated corresponding by combining said pixel with the other pixels of the block encountered along said projection direction or by equality with said pixel when no other pixel is encountered along said projection direction.
    [0007]
    7. System according to claim 6, characterized in that the generated number G ofprojections corresponds to the sum of the minimum number M of projections necessary to reconstruct the generated medium and the number, denoted n, corresponding to the desired level of redundancy.
    [0008]
    8. System according to one of the preceding claims, characterized in that, to generate a bin corresponding to a given combination of pixels, the processing and calculation unit is configured to determine, for the bin to be generated, the combinations of pixels needed; and in that the computing and processing unit (UTC) is configured to generate a program, called a binary generation program, which comprises the following instructions: - reading pixels to be combined with one another from the memory of pixels; generating a bin by combining the values of said pixels with each other; writing the result of the previous operation in the bins memory, or, when there exists a unique relationship between a pixel and a bin to be generated, reading said pixel; generation of the bin from said pixel and writing of said bin in the bins memory, and in that the program is free of instructions for resetting the bin values to be generated, read step, said intermediate read. , the bin to generate and the combining step between a bin to generate and said pixels.
    [0009]
    9. System according to claim 8, characterized in that the processing and calculation unit (UTC) comprises another program, called the bins generation simulation program, including instructions for resetting the bins values to generating, reading, said intermediate reading, the bin to generate and combining a bin to generate and said pixels, and in that the processing and computing unit (UTC) is configured to generate said program for generating bins after executing the bins generation simulation program, the unit being configured to generate said bins generating program by identifying the pixel (s) used by said binary degeneration simulation program to generate the corresponding bins and defining pixel combining instructions which are free from said reset instructions of the bin values to be generated, read step, so-called intermediate reading, the bin to generate and the combining step between the bin to generate and said pixels.
    [0010]
    A data processing method configured to reconstruct a block, called a medium, formed of basic computer data, called pixels, using other computer data, called bins, resulting from a mathematical transform applied to said pixels, a number, denoted G, of groups of bins, called projections, having been generated from the pixels of said block so that said support can be reconstructed with a given minimum number, denoted M, of projections, M being less than G, Characterized in that the method comprises the following steps: a) receiving by a processing and computing unit (UTC) of data, a request to reconstruct the block of pixels; b) reconstruction by the processing and calculation unit (UTC) of a pixel of the support, said univocal pixel, from a bin read in a memory, called bin memory (MB), for which it there exists a unique relation between this bin, called univocal bin, and this univocal pixel, and write in a memory, called pixel memory (MP), distinct from the memory of bins (MB), said unequivocal pixel of the thus calculated block, c ) possibly, reconstruction of other univocal pixels; D) reconstruction using the processing and calculation unit (UTC) of each of the other pixels of the support, said recombination pixels, from the combination of: a bin read in the bins memory ( MB), said combination bin, the generation of which required the application of the multi-pixel transform of the block, with one or more pixels previously reconstructed and read in the pixel memory (MP) and writing into the pixel memory (MP) of each recombination pixel thus calculated.
    [0011]
    11. The method of claim 11, characterized in that the reconstruction 5 of said pixels is performed without writing operation in the bins memory (MB) during the reconstruction of said support pixels.
    [0012]
    Method according to claim 10 or 11, characterized in that said combining operation corresponds to the application of the "exclusive OR" mathematical operator (XOR).
    [0013]
    13. Method according to one of claims 10 to 12, characterized in that for the realization of step d), each pixel to be reconstructed, noted support (k), other than that or those having a unique relationship to a bin , is computed by an equation of the form: support (k) = projection [i] .bins [j] Asupport [] or support (k) = projection [i] .bins [j] Asupport [] ^ ... A support (n) with: 20 k, I, n being indices defining different pixels of the medium to be reconstructed; : the operator XOR projection [i] .bins [j]: the value of a bin of the projection projection [i] that has been generated using the support pixel (k). 25
    [0014]
    14. The method as claimed in claim 13, characterized in that, prior to receiving the reconstruction request, the reconstruction of the pixels is simulated at the level of the processing and calculation unit by executing steps comprising: identification of an unequivocal relationship between a pixel of the medium to be reconstructed and a bin of a projection; - reconstruction of the corresponding pixel in a memory area of the distinct unit of the memory of bins - travels all the bins of at least M projections among the G projections generated to subtract, by application of the XOR operator, the value corresponding to this pixel at each bin generated using said pixel; repetition of the preceding steps for all the pixels of the support; then, from the identified relationships between the pixels and the bins, each pixel to be reconstructed, noted support (k), other than that or those having a unique relationship to a bin, is defined in the processing unit and calculation by an equation of the form: support (k) = projection [i] .bins [j] Asupport [I] or support (k) = projection [i] .bins [j] Asupport [] ^ ... A support (n )
    [0015]
    15. Method according to claim 14, characterized in that the G projections are generated in the following manner: i) - organization of the block of pixels into a rectangle of dimensions {P, Q}, where P is the number of columns and Q is the number of lines, then ii) - definition of a cartesian coordinate system (p, q) whose origin is attached to a pixel preferably forming a corner of the rectangle, iii) - definition of G directions, called projection directions, in the reference (p, q) iv) - for each of said G projection directions, for each pixel of a row or a column of the block, the corresponding bin is calculated by combining said pixel with the other pixels of the block encountered following said projection direction or by equality with said pixel when no other pixel is encountered along said projection direction.
    [0016]
    16. Method according to claim 15, characterized in that the number G of projections generated corresponds to the sum of the minimum number M of projections required to reconstruct the generated medium and the number, denoted n, corresponding to the desired level of redundancy.
    [0017]
    17. Method according to one of claims 10 to 16, characterized in that, to generate a bin corresponding to a given combination of pixels, the necessary combinations of pixels are determined for the bin to be generated; and in that the following steps are performed: - reading pixels to be combined with one another from the pixel memory; generating a bin by combining the values of said pixels with each other; 5 - writing the result of the previous operation in the memory of bins, or, when there is a unique relationship between a pixel and a bin to generate, - reading said pixel; generation of the bin from said pixel and writing of said bin in the bins memory, and in that said bin is generated without steps of resetting the bin values to be generated, read step, said intermediate reading, bin to generate and combining step between a bin to generate and said pixels.
    [0018]
    18. A method according to claim 17, characterized in that, prior to the generation of bins, a generation of bins is simulated which includes instructions for resetting the values of bins to be generated, reading, said intermediate read, the bin to generate and combination between a bin to generate and said pixels, and in that a bins generation program is generated after simulating the generation of bins, said bins generation program being generated by identifying the one or more pixels used in the generation simulation of bins to generate the corresponding bins and defining pixel combination instructions which are free of said reset instructions of the bin values to be generated, read step, said intermediate reading , the bin to generate and the combining step between the bin to be generated and said pixels.
    [0019]
    19. Computer program product downloadable from a communication network and / or stored on a computer readable medium and / or executable by a processor of a processing and computing unit, characterized in that it comprises program code instructions for implementing the steps of a method according to one of claims 10 to 18 when said program is executed by the processor.
    类似技术:
    公开号 | 公开日 | 专利标题
    FR3019338A1|2015-10-02|SYSTEM AND METHOD FOR PROCESSING DATA
    US8983888B2|2015-03-17|Efficient modeling system for user recommendation using matrix factorization
    US9348898B2|2016-05-24|Recommendation system with dual collaborative filter usage matrix
    WO2010076260A1|2010-07-08|Method and system for sorting data from a database
    FR2670920A1|1992-06-26|RESOLUTION OF DENSED SYSTEMS OF LINEAR EQUATIONS ON A COMPUTER WITH PARALLEL PROCESSING.
    WO2018132590A1|2018-07-19|Image recognition method and apparatus based on augmented reality
    EP3639190B1|2021-12-15|Descriptor learning method for the detection and location of objects in a video
    CN108230168B|2021-10-22|Method and system for sharing personalized route planning by means of customized multimedia method
    US20150278910A1|2015-10-01|Directed Recommendations
    WO2020157733A1|2020-08-06|Dynamic three-dimensional imaging method
    US20160259494A1|2016-09-08|System and method for controlling video thumbnail images
    FR2933520A1|2010-01-08|METHOD AND DEVICE FOR RESTORING A VIDEO SEQUENCE
    US20190205757A1|2019-07-04|Model-free control for reinforcement learning agents
    WO2014045094A1|2014-03-27|Hybrid engine for central processing unit and graphics processing unit
    EP1971152A2|2008-09-17|Data management for image processing
    US10750204B2|2020-08-18|System and method for processing data
    BE1021546B1|2015-12-11|METHOD AND SYSTEM FOR STORING WAVEFORM DATA.
    FR3088740A1|2020-05-22|ITERATIVE DATA PROCESSING
    WO2019058052A1|2019-03-28|Method for processing data of a two-dimensional representation of a geographical area
    CN112100046A|2020-12-18|Method and equipment for identifying game similarity
    FR2971601A1|2012-08-17|METHODS, DEVICE AND COMPUTER PROGRAMS FOR RECOGNITION OF FORMS, IN REAL-TIME, USING AN APPARATUS COMPRISING LIMITED RESOURCES
    CN113255709A|2021-08-13|Image element matching method and device, model training method and device and data processing method
    WO2000042575A1|2000-07-20|Method for simplifying a source mesh, taking into account the local curvature and the local geometry, and corresponding uses
    EP2204775B1|2011-10-05|Incremental polygon triangulation for digital display
    FR3098367A1|2021-01-08|Method and device for encoding a sequence of digital holograms
    同族专利:
    公开号 | 公开日
    US20170048551A1|2017-02-16|
    WO2015145078A3|2016-05-06|
    US10182243B2|2019-01-15|
    WO2015145078A2|2015-10-01|
    FR3019338B1|2017-07-14|
    EP3123721A2|2017-02-01|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    CN106204676A|2016-07-12|2016-12-07|大连理工大学|CT method for reconstructing based on Mojette conversion|US6771272B2|2000-03-17|2004-08-03|Sun Microsystems, Inc.|Graphics system having a super-sampled sample buffer with hot spot correction|
    US6816162B2|2002-07-19|2004-11-09|Sun Microsystems, Inc.|Data management to enable video rate anti-aliasing convolution|
    EP2863566B1|2013-10-18|2020-09-02|Université de Nantes|Method and apparatus for reconstructing a data block|
    US9569811B2|2014-06-26|2017-02-14|Qualcomm Incorporated|Rendering graphics to overlapping bins|MX2018001335A|2015-07-31|2019-04-01|Zebware Ab|Data integrity detection and correction.|
    EP3351001B1|2015-09-17|2021-08-11|InterDigital VC Holdings, Inc.|Method for encoding a light field content|
    US10523244B2|2016-08-11|2019-12-31|Zebware Ab|Device and associated methodoloy for encoding and decoding of data for an erasure code|
    US10750204B2|2016-09-16|2020-08-18|Rozo Systems|System and method for processing data|
    US10892781B2|2017-11-20|2021-01-12|Zebware Ab|Method and devices for a reduced repair and update erasure code|
    法律状态:
    2015-09-30| PLFP| Fee payment|Year of fee payment: 3 |
    2016-02-26| TP| Transmission of property|Owner name: ROZO SYSTEMS, FR Effective date: 20160126 |
    2017-12-29| ST| Notification of lapse|Effective date: 20171130 |
    2018-01-16| PLFP| Fee payment|Year of fee payment: 4 |
    2018-02-23| RN| Application for restoration|Effective date: 20180118 |
    2018-03-02| FC| Favourable decision of inpi director general on an application for restauration.|Effective date: 20180129 |
    2018-03-23| PLFP| Fee payment|Year of fee payment: 5 |
    2020-03-19| PLFP| Fee payment|Year of fee payment: 7 |
    2021-03-23| PLFP| Fee payment|Year of fee payment: 8 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1452585A|FR3019338B1|2014-03-26|2014-03-26|SYSTEM AND METHOD FOR PROCESSING DATA|FR1452585A| FR3019338B1|2014-03-26|2014-03-26|SYSTEM AND METHOD FOR PROCESSING DATA|
    US15/126,887| US10182243B2|2014-03-26|2015-03-26|System and method for processing data|
    EP15769349.0A| EP3123721A2|2014-03-26|2015-03-26|System and method for processing data|
    PCT/FR2015/050771| WO2015145078A2|2014-03-26|2015-03-26|System and method for processing data|
    [返回顶部]