Classifications

• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
  • H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
  • H03M7/3059—Digital compression and data reduction techniques where the original information is represented by a subset or similar information, e.g. lossy compression
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
  • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
  • H04N19/136—Incoming video signal characteristics or properties
  • H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
  • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
  • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
  • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
  • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
  • H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
  • H04N19/593—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial prediction techniques
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/65—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using error resilience

Definitions

• the present invention relates to the field of the coding of numerical values arranged in matrix, in particular in the case where this matrix is a two-dimensional matrix, representing the pixels of an image.
• the main constraints of the compression methods are, on the one hand, to reduce as much as possible, by compressing it, the volume, measured in bytes, of an initial digital file and, on the other hand, to restore a file the most as close as possible to the original file.
• Some compression methods make it possible to restore exactly the initial values. This is the case of DPCM modulation.
• an origin value i.e. the first value of the initial digital file
• each other value is replaced by its difference with the value that precedes it in the initial file.
• the numbers corresponding to the differences are generally smaller than those corresponding to the initial values, which makes it possible to obtain a compressed file.
• To restore an initial value it suffices to add the difference corresponding to the previous initial value, that is to say that one adds between them the differences of successive values with the original value.
• This compression mode therefore applies to a linear sequence, that is to say extending in one dimension, of numerical values; the same goes for a two-dimensional matrix of numerical values, consisting of rows and columns, this mode compression that can be applied to each row of the matrix successively; the same applies regardless of the number of dimensions of a matrix, the compression mode that can be applied to each row (or equivalent in the corresponding dimension) of the matrix.
• each image is done in blocks of 8 ⁇ 8 pixels, and in each block, line by line, from the first to the eighth.
• Such a method often results in lineage and / or pixelation of the rendered image.
• the aim of the invention is to propose a simple and powerful compression method which makes it possible to combine the advantages of a greater compression ratio than that of the single DPCM method, while retaining the advantages of a differential coding, without propagation. at the same time as it avoids or significantly reduces the risk of lineage and / or pixelation specific to the processes of the prior art.
• a method for coding successive layers of an initial matrix in a compressed matrix and its restitution in a restored matrix, each cell of the initial matrix containing a respective initial digital value; each cell of the compressed matrix containing a respective compressed digital value corresponding to the respective initial digital value; each cell of the restored matrix containing a respective restored digital value corresponding to the respective initial digital value; characterized in that to process, i.e., compress, at least one initial value of an initial box, one chooses, among the lines passing through said box, as a path of travel for the compression of said initial value, the one which, estimated from returned values previously calculated, has the smallest variation.
• Such a method may comprise at least one layer for which, for an initial value to be processed of said layer, the corresponding compressed value is determined by calculating a difference between said value to be processed and a reference value previously determined and neighboring, preferably the nearest neighbor, the box provided to contain the returned value corresponding to said value to be processed on one of the lines passing through said box.
• the lines passing through the cell intended to contain the returned value corresponding to said value to be processed comprise a value that can potentially be a reference value, for all combinations of potential values taken two by two, for a first of said reference values.
• a variation is calculated equal to the difference between said first value and another previously calculated, restored value located on a same third line that said first value, said third line being parallel to the second line, and said other value being at a same distance from the first value as the second value is from the value to be processed, the selected reference value being the second value for which said variation is the weaker.
• the reference value for which the path gives the smallest variation that is to say that the first reference value is chosen if the second variation is smaller in absolute value than the first variation, or, that the second reference value is chosen if the first variation is smaller in absolute value than the second variation.
• At least one layer preferably a base layer, for which, for an initial value to be processed of said layer, the corresponding compressed value is determined by calculating a difference between the initial value and a reference value equal to one. previously rendered value returned, belonging to either the same row or the same column, as said value to be processed.
• the reference value is calculated from a pair of framing values that have already been processed and disposed of on both sides. another of said value, for which the restored value is calculated from a reference value to the decompression calculated from the same pair of previously restored values.
• the pair whose variation, equal to the difference between the corresponding restituted values of previously restored values, is chosen to calculate the reference value has the lowest absolute value.
• the reference value for this layer is advantageously the arithmetic mean between the two values of the chosen framing torque and, where appropriate, the reference value to the decompression is the arithmetic mean between the values restored of the values of the framing couple;
• the reference value for this layer may alternatively be the arithmetic mean between the two initial values corresponding to that of the chosen framing pair.
• an original initial value whose compressed value and the value restored are equal to said original initial value, said restored original value serving as a reference value, direct or indirect, for the treatment of the other initial values.
• a quantization table is applied to the difference between the initial value processed and the reference value, in order to calculate the compressed value and the value returned.
• the quantization table used being defined from previously computed returned values.
• a threshold below which a first quantization table is applied and beyond which a second quantization table is applied.
• one table or another is used depending on whether the initial value to be treated belongs to one layer or another.
• Each cell can represent a pixel of an image.
• Figure 1 is a partial illustration of an initial matrix, representing the upper right corner of an image, as well as those of the corresponding compressed matrix and error matrix;
• Figures 2 to 5 each illustrate a cell layer among the cells of the initial matrix;
• FIGS. 6 and 7 illustrate two quantization tables respectively for dark and light values of the image, usable during the compression of the values of the base layer
• FIGS. 8 and 9 illustrate two quantization tables respectively for dark and clear values of the image, usable during the compression of the values of the secondary layers.
• FIGS 10 to 14 illustrate algorithms used in the method according to the invention.
• the original image is represented in the figures by an initial matrix M.
• This matrix can be that of the values of a monochrome image, if the original image is monochrome, or one of the matrices making it possible to reconstitute a color image, if the original image is in color.
• the initial matrix M may represent, for each pixel, the brightness of the Red, Green or Blue, in the case of an initial image in RGB notation; it can represent values composed of luminance Y, in the case of an initial image in notation YCbCr.
• the values of the initial matrix M are between 0 and 255.
• the matrix may be the representation of a raw image or of an image that has already undergone a colorimetric transformation.
• line we will call “line”, the lines represented horizontally in the figures and “columns”, the lines represented vertically; of course, whether horizontal or vertical, the lines are treated indifferently according to the method.
• each cell Cmn of the initial matrix M is indexed by its line number m and its column number n and contains a corresponding initial value Vmn, indexed in the same way.
• the initial matrix M corresponds a compressed matrix MC and a matrix MR restored after decompression of the compressed matrix MC.
• each value Vmn of the initial matrix M corresponds a respective compressed value VCmn and a restored value VRmn, each in a respective cell CCmn, CRmn of same indices.
• FIG. 1 thus illustrates the initial matrix M, the compressed matrix MC and the restored matrix MR obtained, starting from the initial matrix M, by the method according to the invention described with reference to FIGS. 10 to 14;
• FIG. 1 also illustrates an error matrix ME consisting of the differences between the restored values and the initial values.
• the method according to the invention allows the successive encoding of successive layers of values, from a Voo origin value contained in a Coo origin cell at the top left of the initial matrix M.
• the encoding of a value of d a given layer only depends on previously encoded layers or previously encoded values of its own layer.
• the method uses four layers, a base layer and three secondary layers.
• the values of the cells belonging to the base layer are represented in bold in the matrix M; in FIG. 3, the values of the cells belonging to the first secondary layer are represented in bold in the matrix M; in FIG. 4, the values of the cells belonging to the second secondary layer are represented in bold in the matrix M; and, in FIG. 5, the values of the cells belonging to the third secondary layer are represented in bold in the matrix M.
• the cells that compose it are regularly distributed in the matrix.
• the base layer is encoded using a first lossless differential encoding algorithm, followed by a first quantization; the secondary layers are encoded using a second predictive algorithm, itself followed by a second quantization.
• Two quantization tables TB1, TB2 are used for quantization of the basecoat values.
• Two other tables TS1, TS2 are used for the quantification of the values of the secondary layers.
• a first table TB1, TS1 is assigned to quantize the dark values; the values being considered as dark below a threshold value equal to 40, out of 255.
• the second tables TB2, TS2 are assigned to the clear values, that is to say from 40 to 255.
• the Choosing the value 40 for the threshold value is arbitrary, determined by a better rendering of the darker colors.
• the first tables TBl and TSl assign to the dark values of the quantifications less severe, corresponding to a division by 2 for the base layer and a division by 4 for the secondary layers, against a division by 3 (TB2) and 5 or 8 (TS2) respectively for clear values.
• Degradations ie errors between the returned value and the initial value are generally more noticeable in dark areas than in bright areas; such an arrangement, allowing to have less losses in the dark areas is particularly advantageous.
• the remainder of the present description explains, with reference to FIGS. 11 to 14, how one determines whether a value is estimated to be light or dark.
• the first column contains a difference to be quantified D
• the second column contains, for each difference D, the corresponding compressed VC value
• the third column contains the corresponding value restored DR of the difference D.
• the number of layers defines a pitch p, q of the base layer, i.e. a distance between the cells belonging to the first layer.
• the pitch p along the lines and the pitch q according to the columns are equal, with:
• the cells of the base layer have as indices:
• n 4 z, with z> 0 and,
• the base cells thus constitute a mesh among the cells of the initial matrix M, from which the compressed values of the other cells are calculated.
• an encoding that is little or no destructive relative to the encoding of the other cells thus, regularly distributed values throughout the image are restored sufficiently close to the initial values to maintain a quality impression. while being able to afford greater compression of the values of the secondary layers.
• the upper left corner shown corresponds to a mesh; with for each of the cells represented:
• FIGS. 10 to 14 Each of these figures represents the computation of a compressed value VCmn, ie a step in the constitution of the compressed matrix MC, especially in function of the initial values Vmn, of the initial matrix M, but also as a function of VR restored values of previously compressed values.
• Figures 11 and 12 each illustrate the compression of a respective value among those of the base layer cells.
• Figures 13 and 14 each illustrate the compression of a respective value among those of the cells of the first secondary layer.
• the compression / reproduction method according to the invention retains the value Voo of the Coo origin cell, identified in FIG. 10, so that in the illustrated example, we have:
• the VRoo rendered origin value is a reference, directly or indirectly, for the calculation of the other compressed values VCmn or restored VRmn.
• the Co4 cell belongs to the base layer. Apart from the special case of the Voo value, the value V04 is the first to be compressed.
• the value returned, already calculated, of the nearest lower index on the same row or column is taken as the reference value.
• the reference value being less than 40 the value Vo4 is treated as a dark value and therefore the first table TB1 is used.
• the same procedure is applied to the initial value V4o of the cell C4o of the initial matrix M.
• the same procedure can then be applied to all the initial values Vo, 4z and V4y, o.
• the C44 cell belongs to the base layer.
• the value returned, already calculated, of the nearest lower index on the same row or column is taken as the reference value.
• two values answer potentially to this definition, one VRo4 in the box CRo4, the other VR4o in the box CR4o of the matrix MR.
• the application of the method according to the invention to cells of a secondary layer of the initial matrix M will now be described with reference to FIGS. 13 and 14.
• the cells of each secondary layer are all surrounded by cells of the lower index layer (in the case of FIG. 13), or by cells of the same layer themselves framed by cells of the lower index layer (the case of FIG. 14),
• a reference value is chosen as an average value between the values of a framing pair; a framing pair being a pair of previously computed returned values contained in two adjacent cells, on either side of the cell being processed, in the restored matrix MR, aligned on the same line m or on the same column not.
• This average is in this example rounded to the nearest integer.
• the compression reference value W22 is then calculated from the original values located in the equivalent cells:
• the restituted values are used to calculate the following restituted values, to determine, if necessary, the direction of the decompression for each value to be restored, and to determine the quantization table that will be used as a function of the chosen threshold value.
• the compressed matrix MC thus contains in itself all the information necessary for the restitution.
• Such an arrangement can also be advantageously used for a video game, particularly for a game played "online", in which fast sessions can be chained at high frequency; a treatment of only one zone of the image, or a slowing down of the action, would be detrimental to the pleasure of the player, to the success of his actions; the treatment of successive layers distributed in the image, each layer being transmitted and displayed as soon as it is processed, avoids these disadvantages.
• the corresponding compressed values can be calculated from the average values restored with the same mn indices as the framing pair, and the values returned from these same values.
• the origin cell may be any other cell of the initial matrix, so that the row and / or column numbers may take negative values, the compression principle according to the invention is not affected.
• a compression method according to the invention is applicable to any type of matrix, in particular to that corresponding to large amounts of data, for example statistical data.
• the invention is applicable to matrices of greater dimensions; in particular, a video can be represented by a three-dimensional matrix, the third dimension representing time.
• Such a compression method is, moreover, particularly suitable for use in the processing of massive data, for example meteorological data, which can be represented in the form of a matrix with two or more dimensions.
• the base layer has its own quantization tables and the quantization tables used for the compression of the secondary layers are identical for all the secondary layers.
• the quantization tables used for the compression of the secondary layers are identical for all the secondary layers.
• the order of processing of the initial values in each layer can be changed. They can be processed by rows or by successive columns, or in another order. It can also be expected that a layer treatment will begin before the previous layer treatment is completed.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Theoretical Computer Science (AREA)
• Compression Or Coding Systems Of Tv Signals (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)
• Compression Of Band Width Or Redundancy In Fax (AREA)
• Organic Low-Molecular-Weight Compounds And Preparation Thereof (AREA)
• Investigating Or Analysing Biological Materials (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=48170541

Family Applications (1)

Country Status (9)

Families Citing this family (4)

Family Cites Families (8)

• 2012
  • 2012-10-05 FR FR1259484A patent/FR2996706A1/fr active Pending
• 2013
  • 2013-10-07 EP EP13801611.8A patent/EP2904707A1/fr not_active Withdrawn
  • 2013-10-07 CA CA2887366A patent/CA2887366C/en not_active Expired - Fee Related
  • 2013-10-07 WO PCT/FR2013/052369 patent/WO2014053791A1/fr not_active Ceased
  • 2013-10-07 US US14/433,224 patent/US9628806B2/en not_active Expired - Fee Related
  • 2013-10-07 BR BR112015007461A patent/BR112015007461A8/pt not_active IP Right Cessation
  • 2013-10-07 JP JP2015535097A patent/JP6367202B2/ja not_active Expired - Fee Related
  • 2013-10-07 KR KR1020157011903A patent/KR102067612B1/ko not_active Expired - Fee Related
  • 2013-10-07 IN IN3760DEN2015 patent/IN2015DN03760A/en unknown

Non-Patent Citations (2)

Also Published As

Similar Documents

Legal Events