Patents by Inventor Christophe Gisquet

Christophe Gisquet has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Patent number: 10469842
    Abstract: The invention relates to lossless encoding of pixel blocks. To apply the Palette coding mode, an input 4:2:0 block is up-sampled into a 4:4:4 block. Each 4:4:4 pixel is usually lossless encoded using a fully matching entry from the palette, thus being sub-optimal in terms of rate. This is mainly because lossless encoding is still used even for Chromas that have no impact on the decoded output image. To improve this situation, the invention provides that encoding the up-sampled 4:4:4 block includes encoding a 4:4:4 pixel corresponding to a 4:2:0 input pixel without Chromas, using a palette entry different from the 4:4:4 pixel. A lossy matching counter-intuitively does not provide losses in the decoded output image, since it is limited to Chromas discarded in the 4:2:0 block. Less palette entries are thus required, saving indexes and thus reducing the encoding rate.
    Type: Grant
    Filed: February 6, 2017
    Date of Patent: November 5, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche, Patrice Onno
  • Patent number: 10462493
    Abstract: The present invention provides a method and a device for coding and decoding at least a part of an image comprising a plurality of samples, each sample comprising at least two components, and the encoded image comprising at least one filtering parameter. The filtering comprises using at least one filtering parameter as a common filtering parameter for filtering both the first component and the second component of a reconstructed sample.
    Type: Grant
    Filed: June 7, 2017
    Date of Patent: October 29, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Guillaume Laroche, Christophe Gisquet, Edouard Francois, Patrice Onno
  • Patent number: 10448021
    Abstract: For an image portion to encode an initial set of motion information predictors is obtained. It is tested whether the number of motion information predictors in the initial set is lower than a target number and, if so, one or more motion information predictors are added to the initial set to generate a set of motion information predictors having controlled diversity. The motion information predictors of the initial set are actual motion information predictors, having motion vectors obtained from image portions of the image being encoded or of a reference image, and potential motion information predictors for addition include one or more further such actual motion information predictors and also include one or more virtual motion information predictors not having motion vectors obtained from image portions of said image being encoded or of a reference image. A motion information predictor for the image portion to encode is selected from the generated set of motion information predictors.
    Type: Grant
    Filed: August 8, 2018
    Date of Patent: October 15, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche
  • Patent number: 10425645
    Abstract: The invention relates to the encoding of pixel blocks. To apply some encoding tools, an input 4:2:0 block is upsampled into a 4:4:4 block. Each color component of the 4:4:4 block has thus the same importance when evaluating a distortion measure. To vary the influence of the color components depending on their impact on the decoding, the invention provides assigning weights to the color components of pixels of the 4:4:4 block depending on whether the corresponding color components exist in the 4:2:0 block, due to the subsampling. In such a case, any measure is evaluated using the weighted color components. This approach may apply to the palette coding mode, for instance when updating the color component values of a palette entry upon building, or selecting the closest palette entry to a pixel currently considered, or also determining whether a palette entry predictor may be used instead of a palette entry.
    Type: Grant
    Filed: September 29, 2016
    Date of Patent: September 24, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche, Patrice Onno
  • Patent number: 10419774
    Abstract: A method for determining encoding parameters for encoding indexes of a block of indexes used for encoding a coding unit of an image according to a palette, each index being related to one of the pixels forming the coding unit, the palette comprising a set of indexes associated to pixel values. The method comprises: generating each index of the block of indexes related to the coding unit from the palette based on corresponding encoding parameters, and if indexes of the block of indexes cannot be generated from the palette for at least one pixels of the coding unit, generating index values corresponding to escape values for replacing said indexes of the block of indexes which cannot be generated, wherein encoding parameters corresponding to an index are determined according to the spatial neighborhood of said pixel regarding to an encoding order.
    Type: Grant
    Filed: December 10, 2014
    Date of Patent: September 17, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Guillaume Laroche, Christophe Gisquet, Patrice Onno
  • Publication number: 20190281322
    Abstract: The present invention provides a method and a device for coding and decoding at least a part of an image comprising a plurality of samples, each sample comprising at least two components, and the encoded image comprising at least one filtering parameter. The filtering comprises using at least one filtering parameter as a common filtering parameter for filtering both the first component and the second component of a reconstructed sample.
    Type: Application
    Filed: May 30, 2019
    Publication date: September 12, 2019
    Inventors: Guillaume Laroche, Christophe Gisquet, Edouard Francois, Patrice Onno
  • Patent number: 10382765
    Abstract: A video signal has at least a first signal component corresponding to a first color, and a second signal component, associated with the first signal component and corresponding to a second color. Each signal component is divided into blocks, and each block has one or more encodable units. The encoding of at least one encodable unit of the second signal component is switchable between first and second modes. A method of encoding the signal comprises excluding, for at least one encodable unit of the second signal component, use of the second mode for the encoding of the second-signal-component encodable unit concerned when at least one predetermined condition is satisfied relating to an encoded block or encodable unit of the first signal component that corresponds to the second-signal-component encodable unit concerned.
    Type: Grant
    Filed: June 28, 2013
    Date of Patent: August 13, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Edouard Francois, Guillaume Laroche, Patrice Onno, Christophe Gisquet
  • Patent number: 10368106
    Abstract: The present invention provides a method and a device for coding and decoding at least a part of an image comprising a plurality of samples, each sample comprising at least two components, and the encoded image comprising at least one filtering parameter. The filtering comprises using at least one filtering parameter as a common filtering parameter for filtering both the first component and the second component of a reconstructed sample.
    Type: Grant
    Filed: June 7, 2017
    Date of Patent: July 30, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Guillaume Laroche, Christophe Gisquet, Edouard Francois, Patrice Onno
  • Publication number: 20190230369
    Abstract: The invention relates to video coding. A method of encoding or decoding such a video sequence according to the invention comprises: obtaining a reconstructed frame area made of an array of sample values; applying a sample adaptive offset, SAO, filtering to the reconstructed frame area based on SAO parameters; wherein at least one of the SAO parameters is determined independently of the sample values of the reconstructed frame area. Preferably, the sample-independent SAO parameters define an Edge Offset filtering.
    Type: Application
    Filed: March 29, 2019
    Publication date: July 25, 2019
    Inventors: GUILLAUME LAROCHE, EDOUARD FRANÇOIS, CHRISTOPHE GISQUET, PATRICE ONNO
  • Patent number: 10349056
    Abstract: For an image portion to encode an initial set of motion information predictors is obtained. It is tested whether the number of motion information predictors in the initial set is lower than a target number and, if so, one or more motion information predictors are added to the initial set to generate a set of motion information predictors having controlled diversity. The motion information predictors of the initial set are actual motion information predictors, having motion vectors obtained from image portions of the image being encoded or of a reference image, and potential motion information predictors for addition include one or more further such actual motion information predictors and also include one or more virtual motion information predictors not having motion vectors obtained from image portions of said image being encoded or of a reference image. A motion information predictor for the image portion to encode is selected from the generated set of motion information predictors.
    Type: Grant
    Filed: August 8, 2018
    Date of Patent: July 9, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche
  • Publication number: 20190191162
    Abstract: The present invention concerns a method for encoding a video sequence, comprising the following steps by a processing unit of an encoding device: splitting an image from the video sequence into blocks of values; for each block: transforming the values of the block into transform coefficients; organizing these transform coefficients into several sets of transform coefficients; quantizing the transform coefficients into quantized coefficients; encoding the block using the quantized coefficients; encoding the video sequence based on the encoding of the blocks; ?wherein the quantizing step further comprises for quantized coefficients corresponding to one set of transform coefficients: determining scaled coefficients where a scaled coefficient is based on a transform coefficient of the one set, by multiplying the transform coefficient by a factor parameter and then adding a deadzone parameter, the deadzone parameter being scaled linearly with the index k of the transform coefficient within the set of transf
    Type: Application
    Filed: November 30, 2018
    Publication date: June 20, 2019
    Inventors: Christophe Gisquet, Patrice Onno, Guillaume Laroche
  • Publication number: 20190191161
    Abstract: The present invention concerns a method for encoding a video sequence, comprising the following steps by a processing unit of an encoding device: splitting a digital image from the video sequence into blocks of values; for each block: transforming the values of the block into transform coefficients; organizing these transform coefficients into several sets of transform coefficients; quantizing the transform coefficients into quantized coefficients; encoding the block using the quantized coefficients; encoding the video sequence based on the encoding of the blocks; wherein the quantizing step further comprises for quantized coefficients corresponding to one set of transform coefficients: comparing a sum value representing a result of summing magnitudes of the quantized coefficient values with a threshold, the threshold depending on the number of the summed quantized coefficients, the quantized coefficient magnitudes being comprised within a predefined range; and setting the quantized coefficients to zer
    Type: Application
    Filed: November 30, 2018
    Publication date: June 20, 2019
    Inventors: Christophe Gisquet, Patrice Onno, Guillaume Laroche
  • Patent number: 10313668
    Abstract: The present invention concerns a method for encoding an image, comprising the encoding of mode information representing a prediction mode for encoding a current block of pixels according to an Intra coding mode, the method comprising for the determination of set of most probable modes defining at least one candidate for a most probable mode to be the mode information of a neighbor-block if it exists; wherein a candidate for a most probable mode is the horizontal or vertical mode information if this neighbor-block exists and has been encoded using a predetermined coding mode.
    Type: Grant
    Filed: January 27, 2016
    Date of Patent: June 4, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche, Patrice Onno
  • Patent number: 10298937
    Abstract: The invention relates to video coding. A method of encoding or decoding such a video sequence according to the invention comprises: obtaining a reconstructed frame area made of an array of sample values; applying a sample adaptive offset, SAO, filtering to the reconstructed frame area based on SAO parameters; wherein at least one of the SAO parameters is determined independently of the sample values of the reconstructed frame area. Preferably, the sample-independent SAO parameters define an Edge Offset filtering.
    Type: Grant
    Filed: January 2, 2014
    Date of Patent: May 21, 2019
    Assignee: CANON KABUSHIKI KAISHA
    Inventors: Guillaume Laroche, Edouard François, Christophe Gisquet, Patrice Onno
  • Patent number: 10277894
    Abstract: The invention concerns a method and a device for processing images, e.g. for encoding or decoding the images into or from a bitstream, using a palette prediction mode as defined in HEVC SCC. The images are split into hierarchical coding structures.
    Type: Grant
    Filed: January 29, 2016
    Date of Patent: April 30, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Christophe Gisquet, Guillaume Laroche, Patrice Onno
  • Publication number: 20190098314
    Abstract: For an image portion to encode an initial set of motion information predictors is obtained. It is tested whether the number of motion information predictors in the initial set is lower than a target number and, if so, one or more motion information predictors are added to the initial set to generate a set of motion information predictors having controlled diversity. The motion information predictors of the initial set are actual motion information predictors, having motion vectors obtained from image portions of the image being encoded or of a reference image, and potential motion information predictors for addition include one or more further such actual motion information predictors and also include one or more virtual motion information predictors not having motion vectors obtained from image portions of said image being encoded or of a reference image. A motion information predictor for the image portion to encode is selected from the generated set of motion information predictors.
    Type: Application
    Filed: November 26, 2018
    Publication date: March 28, 2019
    Inventors: Christophe Gisquet, Guillaume Laroche
  • Publication number: 20190098313
    Abstract: For an image portion to encode an initial set of motion information predictors is obtained. It is tested whether the number of motion information predictors in the initial set is lower than a target number and, if so, one or more motion information predictors are added to the initial set to generate a set of motion information predictors having controlled diversity. The motion information predictors of the initial set are actual motion information predictors, having motion vectors obtained from image portions of the image being encoded or of a reference image, and potential motion information predictors for addition include one or more further such actual motion information predictors and also include one or more virtual motion information predictors not having motion vectors obtained from image portions of said image being encoded or of a reference image. A motion information predictor for the image portion to encode is selected from the generated set of motion information predictors.
    Type: Application
    Filed: November 26, 2018
    Publication date: March 28, 2019
    Inventors: Christophe Gisquet, Guillaume Laroche
  • Publication number: 20190098333
    Abstract: The invention relates to the encoding or decoding of at least one portion of an image by predicting the at least one portion using at least one predictor, the at least one predictor being determined as a function of at least one reference image portion and as a function of values of a plurality of items of motion information. After having obtained a value of a first item of the plurality of items, a value of a second item is evaluated as a function of the obtained value of the first item, the second item being an item of the plurality of items and being distinct from the first item. Next, the at least one predictor is determined as a function of the at least one reference image portion and as a function of the first and second items.
    Type: Application
    Filed: November 28, 2018
    Publication date: March 28, 2019
    Inventors: GUILLAUME LAROCHE, CHRISTOPHE GISQUET, PATRICE ONNO
  • Publication number: 20190098315
    Abstract: For an image portion to encode an initial set of motion information predictors is obtained. It is tested whether the number of motion information predictors in the initial set is lower than a target number and, if so, one or more motion information predictors are added to the initial set to generate a set of motion information predictors having controlled diversity. The motion information predictors of the initial set are actual motion information predictors, having motion vectors obtained from image portions of the image being encoded or of a reference image, and potential motion information predictors for addition include one or more further such actual motion information predictors and also include one or more virtual motion information predictors not having motion vectors obtained from image portions of said image being encoded or of a reference image. A motion information predictor for the image portion to encode is selected from the generated set of motion information predictors.
    Type: Application
    Filed: November 26, 2018
    Publication date: March 28, 2019
    Inventors: Christophe Gisquet, Guillaume Laroche
  • Patent number: 10178405
    Abstract: A decoding method wherein a motion vector is associated with a block of pixels of a current image, wherein the image is subdivided into spatial subdivisions, wherein the block of pixels belongs to a current spatial subdivision and is to be predicted using an Intra Block Copy mode, using a predictor block of pixels of the image designated by the motion vector, and wherein the predictor block designated by the motion vector may belong to any one of previously decoded spatial subdivisions in the current image. Embodiments provide enhancements to the Intra Block Copy Mode of the HEVC standard.
    Type: Grant
    Filed: June 3, 2016
    Date of Patent: January 8, 2019
    Assignee: Canon Kabushiki Kaisha
    Inventors: Guillaume Laroche, Christophe Gisquet, Patrice Onno