Classifications

• • 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
• • 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
  • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
  • H04N19/103—Selection of coding mode or of prediction mode
• • 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
  • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
  • H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
• • 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
  • H04N19/146—Data rate or code amount at the encoder output
  • H04N19/147—Data rate or code amount at the encoder output according to rate distortion criteria
• • 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
  • H04N19/189—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding
  • H04N19/19—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding using optimisation based on Lagrange multipliers
• • 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
  • H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
  • H04N19/36—Scalability techniques involving formatting the layers as a function of picture distortion after decoding, e.g. signal-to-noise [SNR] scalability
• • 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
  • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
  • H04N19/507—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction using conditional replenishment

Definitions

• the present invention relates to a method and a device for compressing image parts. It applies, in particular to the coding of images and sequences of images, in particular for transmission over a computer network, for example the Internet, or storage on a digital information medium.
• This method and device can be integrated into any system for compressing and decompressing any image portion on any hardware platform.
• FIG. 1 represents, in the form of a functional block diagram, a device implementing such a method.
• a prediction, performed by the function 102, of a portion of an original image 101 is established from other parts of one or more images subsequent to or prior to the image being encoded, or other elements of the current image.
• the residue 103 is obtained by the difference 105 between the prediction and the original.
• the residue 103 and the prediction 102 are encoded in the transmitted video stream by the encoding function 104.
• This method is typically used to exploit the temporal redundancy of a sequence of images. It is more particularly used in the ISO / ITU MPEG and H.261 / 3/4 Standards compression system. As illustrated in FIG.
• motion compensation is based on the comparison of a given block of the current image with a set of blocks or sub-blocks of the current image or other images posterior to or before the current image. A comparative measurement is established between the blocks. If the difference measurement is below a certain threshold, the blocks are considered as similar and the position difference is indicated by a motion vector.
• the known video compression standards such as MPEG, use the two groups of steps 200 and 210, described with reference to FIG. 2, to compress the images of a video stream into non-intra images, ie images. which are encoded by reference to other images in the video stream.
• the motion estimation 200 (in English "motion estimation") consists in encoding each image from elements present in other images of the video stream, called reference images (often the previous image, the last image intra or the next intra picture). The steps performed during motion estimation 200 are as follows:
• step 201 cutting of the image to be encoded into blocks of fixed size L x H pixels (often 16 x 16);
• step 202 for each block of the image, search in the reference image of the floating area of L x H pixels most similar to the block considered;
• step 203 for each block, storage of the displacement vector
• motion vector which indicates the displacement (in spatial coordinates) between the block and the most similar zone found during step 202;
• Motion compensation 210 (in English "motion compensation") consists of compressing the residue. The steps performed during the motion compensation 210 are, for each block, the following:
• step 214 for each block, calculation of the residue, that is to say of the difference between the block and the most similar zone found.
• step 21 1 compression of the residue, most often through a mathematical function, such as OCT (acronym for "discrete cosine transform” for discrete cosine transform);
• step 212 storage of the compressed residue
• step 213 return to step 200 to encode the next block, if it remains to be processed.
• step 221 decompression of one or more intra images (encoded images without reference to other images of the video stream);
• step group 230 reconstitution of non-intra images by performing, for each block:
• step 231 locating, thanks to the displacement vector, the most similar zone of another image of the video stream;
• step 232 decompression of the residue
• step 233 adding the residue to the most similar zone to obtain the final block and returning to step 230 for the next block, if there are still any to be processed.
• the blocks of the predicted images are predicted from parts of other images and decoded motion vectors, to which must be added the coefficients of the motion compensation.
• the choice of motion vector can be made according to different criteria “SAD” (sums of the absolute differences) or “SSD” (sums of the squares of the differences) which take account only of the generated distortion.
• SAD sums of the absolute differences
• SSD sums of the squares of the differences
• a “rate distortion optimization” optimization (known by the acronym “RDO” in English and meaning optimization of the bit rate and the distortion in French) is used and takes into account the flow generated to transmit the information (size taken in the flow) and the distortion provided by this information (relevance of the information)
• the purpose of the RDO is to find a compromise between the bit rate used by the coded stream and the relevance of the information to be coded.
• the RDO algorithm can be used during motion compensation.
• the motion estimation step between two images provides several candidate motion vectors.
• Each of the motion vectors is then transformed in the same way as if it had been transformed to be encoded in the video stream, and then decoded in the same way as if it had been decoded from the video stream.
• the distortion between the original block and the coded and decoded block, calculated after decoding the motion vector, is determined for each candidate motion vector.
• the motion vector that minimizes distortion and throughput is retained.
• Patent Application EP0866622 entitled “Video Oata compression method” defines a performance measure (which takes into account the coding parameters 8) that is applied to the motion compensation step in order to improve the measurement of costs. in flow and distortion.
• the RDO algorithm makes it possible to establish a flexible decision criterion as to the choice of coding criteria for the coefficients.
• a tree is created whose nodes represent the various coding steps and measure the costs in bit rate and distortion generated. The tree is then traversed by taking the optimal path (minimizing the flow and distortion costs) to obtain the most relevant coding choice for the rate and distortion generated by the coding of the coefficients.
• the motion compensation step is eliminated.
• the visual rendering is significantly deteriorated.
• the present invention aims, in a first aspect, a method of compressing parts of images which comprises:
• the invention optimizes the size of the video stream and the resources required for decompression by coding in the video stream only the most relevant data.
• the residue is not encoded in the stream.
• the decoding of the coefficients of the residue and the addition of the residual to the reconstructed block which are costly in terms of resources, are avoided.
• the determination of the distortion related to the single code of each prediction is accurate since it has, at the coding level, the portion of the image as it would be decoded in the absence of motion compensation.
• the process as briefly described above also comprises:
• a residue determination step the difference between a prediction representative of each prediction and the image portion to be compressed
• a step of determining distortion associated with the residue said decision step implementing said flow rate associated with the residue and the distortion associated with the residue.
• the process as briefly described above also comprises:
• the determination of the distortion related to the coding of the residue is accurate since it has, at the coding level, the residue that would be obtained at the decoding level.
• Rr is the flow rate associated with the residue
• k is the product of Rp and ⁇ . It does not need to be determined to make the comparisons below since the term is present in RDI and RD2.
• - Rp is the rate associated with each prediction
• - Dp is the distortion associated with each prediction
• - ⁇ is a positive predetermined coefficient. if RDI ⁇ RDs, only each prediction is coded and if RD 1> RDs, each prediction and the residue are coded.
• the coder is simplified and the resources consumed in encoding are limited since the decision to code the residue depends only on a distortion rate criterion taking into account only the bit rates and distortions related to each prediction.
• - ⁇ is a positive predetermined coefficient. if RDI ⁇ RDs, only each prediction is coded and if RD 1> RDs, each prediction and the residue are coded.
• the coder is simplified and the resources consumed are limited to the coding since the decision to code the residue depends only on a distortion criterion.
• - Dp is the distortion associated with each prediction
• - ⁇ is a positive predetermined coefficient
• k is the product of Rp and ⁇ . It does not need to be determined to make comparisons between RDI and RD2 since the term is present in RDI and RD2, and is optional for comparisons between RDI and RDs.
• RD2 k + Rr * ⁇ + Dr: where: - Rr is the flow rate associated with the residue,
• the coder is simplified and the resources consumed in the coding are limited since the decision to code the residue depends only on a distortion rate criterion taking into account only the distortion and possibly the bit rates related to each prediction. .
• the present invention provides a device for compressing an image part, which comprises:
• FIG. 1 represents, in the form of a functional block diagram, a coding method with coding of the prediction and of its residue known in the prior art
• FIG. 2 represents, in the form of a logic diagram, steps implemented in a compression method and in an associated image decompression method, known in the prior art
• FIGS. 3 to 6 represent, in the form of logic diagrams, steps implemented in particular embodiments of the method that is the subject of the present invention.
• FIG. 7 represents a particular embodiment of the device that is the subject of the present invention.
• the image to be encoded is broken down into blocks and stored in a memory zone of a coding or compression device.
• the following steps, 302 to 316, are performed, successively, for each block of the image to be encoded.
• a motion estimation is carried out in a known manner, for example as described in the patent application FR2850827 to provide a prediction, with respect to each reference image.
• the distortion caused by each encoded prediction is determined and stored in memory, and optionally, the rate corresponding to each prediction is determined and stored in memory.
• a measure of the distortion corresponding to the block obtained during step 312, is then calculated, for example by the SSD method.
• This distortion corresponds, as well as the distortion that would be generated by the decoding, with the single prediction (s) encoded in the video stream.
• step 303 the difference between the prediction and the original block to be coded is determined, and the residue is written in memory, that is to say the result of this difference.
• the residue is transformed into the frequency domain. Then, in a step 309, the transform of the residue is quantized and the quantized transformed residue is coded.
• step 310 the inverse quantization of the transformed quantized residue is carried out.
• step 311 the transform in the spatial domain of the result of step 310 is performed.
• the result thus obtained is the residue that the decoding device would decode from the coded video stream.
• step 313 the rate required for the encoding of the transformed and quantized residue resulting from step 309 is determined and the distortion generated by the decoded residue resulting from step 311 is calculated. These are stored in memory. flow and distortion.
• step 314 it is determined whether, for the block considered, the result of step 309 must be transmitted to the decoding device. For this purpose, from the data in memory, the following decision parameters RDI and RD2 are calculated:
• Rp is the flow rate of the predicted block
• - Dp is the distortion of the predicted block
• - Rr the flow of the residue
• k is the product of Rp and ⁇ . It is not necessary to determine it for the comparisons below since the term is present in RDI and RD2.
• each prediction is coded in the video stream, for example by means of the coding of its or its motion vectors.
• RDI the coding of the motion compensation data is relevant to the bit rate and gain in distortion costs.
• each prediction and the motion compensation are coded in the video stream, for example by means of the coding of its or its motion vectors and of the transformed and quantized residue.
• RDI represents a rate-distortion criterion related solely to the coding of each prediction and RD2 represents a rate-distortion criterion related to the coding of each prediction and of the residue.
• the factor representing the flow rate is (Rp + Rr)
• the step of determining the Rp, rate associated with each prediction is not necessary since the term "Rp * ⁇ " is common to both decision parameters RDI and RD2. It is therefore possible to compare RDI and RD2 without calculating k.
• the image to be coded in blocks is decomposed and the blocks of the image to be encoded are stored in a memory zone. The following steps, 402 to 416, are performed, successively, for each block of the image to be encoded.
• a motion estimation is performed and at least one prediction is provided.
• step 405 the distortion Dp generated by the prediction is determined.
• the rate Rp required to send each prediction is determined.
• step 403 the difference between the prediction and the corresponding block of the original image to be encoded is determined and this difference, called “residue”, is stored.
• a frequency transformation of the residue is carried out by a DCT type transformation, and a quantification of the residue transform is carried out.
• the Rr rate necessary for the encoding of the transformed and quantized residue and the distortion Dr that the coded residue generates are determined and stored in memory.
• Rp is the flow rate of the predicted block
• k is the product of Rp and ⁇ . It does not need to be determined to make the comparisons below since the term is present in RDI and RD2.
• RDI represents a rate-distortion criterion related solely to the coding of each prediction and RD2 represents a rate-distortion criterion related to the coding of each prediction and of the residue.
• the factor representing the flow rate is (Rp + Rr), one can thus take into account the two flow rates: that of the predicted block and that of the residue, since the residue is necessarily encoded with the predicted block.
• the step of determining Rp, the rate associated with each prediction is not necessary since the term "Rp * ⁇ " is common to the two decision parameters RDI and RD2. It is therefore possible to compare RDI and RD2 without determining k. If RDI ⁇ RD2, the coding of the motion compensation data is not relevant to the bit rate and distortion gains, and only each prediction is encoded in the video stream, during the step 416.
• the encoding of the motion compensation data is relevant to the bit rate and distortion gains, and in a step 415 the predicted block and the motion compensation are encoded. in the video stream.
• the image to be coded in blocks is decomposed and the blocks of the image to be encoded in memory are stored.
• the following steps 502 to 516 are performed, successively, for each block of the image to be encoded.
• the motion estimation of the current block of the image to be coded is performed and at least one prediction is provided.
• step 503 the difference between the prediction and the block of the original image to be coded is made and the result of this difference, or residue, is written in memory.
• step 514 it is estimated, with tables, the Dp distortion and optionally the Rp rate that would generate the coding of each prediction in the video stream and stores Dp and optionally Rp in memory, it is estimated, with tables , the rate Rr and the distortion Dr that would generate the coding of the residue in the video stream and stores Rp and Dp in memory.
• step 514 it is determined whether, for the block in question, the residue must be transmitted to the decoding device.
• Rp is the flow rate of the predicted block
• RDI represents a rate-distortion criterion related solely to the coding of each prediction
• RD2 represents a rate-distortion criterion related to the coding of each prediction and of the residue.
• the factor representing the flow rate is (Rp + Rr)
• the two flow rates that of the predicted block and that of the residue, since the residue is necessarily encoded with the predicted block.
• the step of determining the Rp, rate associated with each prediction is not necessary since the term "Rp * ⁇ " is common to both decision parameters RDI and RD2. It is therefore possible to compare RDI and RD2 without determining k.
• the coding of the motion compensation data is not relevant to the bit rate and distortion gains, and only each prediction is encoded in the video stream, during the step 516.
• the coding of the motion compensation data is relevant to the bit rate and distortion gains, and in a step 515 the predicted block and the motion compensation are encoded. in the video stream.
• the image to be encoded is broken down into blocks and stored in a memory zone during a step 601.
• the following steps 602 to 616 are performed, successively, for each block of the image to code.
• the motion estimation of the block of the image to be encoded is performed and at least one prediction is provided.
• the rate Rp and optionally the distortion are calculated.
• Dp would generate the coding of each prediction in the video stream and store Rp and optionally Dp in memory.
• - Dp is the distortion of the predicted block and - ⁇ , positive, is set by the designer or the user of the encoder / decoder or is set according to the resources available at the decoding device.
• RDI RDs
• RDs is a threshold value determined by the designer of the encoder
• FIG. 7 shows a particular embodiment of a device for compressing image portions 705, object of the present invention.
• This device 705 includes an image representative signal input 725, a processor 710, a program memory 715, an image memory 720 and a compressed image signal output 730.
• the processor 710 is of known type. In combination with the memories 715 and 720, it is adapted to implement an embodiment of the method that is the subject of the present invention, for example one of those illustrated in FIGS. 3 to 6.
• the program memory 715 contains instructions readable by the processor 710 and implementing the steps of the method to be implemented.
• the processor 710 constitutes, at least:
• a prediction of a block can be made from a set of blocks previously processed and belonging to the same image as the block being processed. This prediction can be done by means of a function applied to this set of blocks, for example the function can be a combination of the colorimetric information of the blocks adjacent to the block being processed.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Compression Or Coding Systems Of Tv Signals (AREA)
• Compression Of Band Width Or Redundancy In Fax (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=38293927

Family Applications (1)

Country Status (5)

Families Citing this family (3)

Family Cites Families (18)

• 2006
  • 2006-10-27 FR FR0609432A patent/FR2907989B1/fr active Active
• 2007
  • 2007-10-26 US US12/446,835 patent/US8532179B2/en active Active
  • 2007-10-26 JP JP2009533739A patent/JP5468383B2/ja active Active
  • 2007-10-26 EP EP07819360A patent/EP2084910A1/de not_active Ceased
  • 2007-10-26 WO PCT/EP2007/009315 patent/WO2008049628A1/fr not_active Ceased
• 2013
  • 2013-08-13 US US13/966,008 patent/US9628791B2/en not_active Expired - Fee Related

Non-Patent Citations (3)

Also Published As

Similar Documents

Legal Events