Patents by Inventor Rajan Laxman Joshi

Rajan Laxman Joshi 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: 10284849
    Abstract: Techniques for quantization parameter (QP) for display stream compression (DSC) based on complexity measure are disclosed. In one aspect, a method for determining a QP value includes determining a complexity value of a plurality of previous blocks and selecting a technique from a plurality of defined techniques for calculating a QP adjustment value for a current block based on the determined complexity value. The method may further include calculating the QP adjustment value for the current block via the selected technique and determining the QP value for the current block based on the QP adjustment value.
    Type: Grant
    Filed: April 6, 2016
    Date of Patent: May 7, 2019
    Assignee: QUALCOMM Incorporated
    Inventors: Vijayaraghavan Thirumalai, Natan Haim Jacobson, Rajan Laxman Joshi
  • Patent number: 10277915
    Abstract: The techniques of this disclosure may be generally related to signaling values of a quantization matrix. In some examples, coefficient values in the quantization matrix may be downsampled with different factors based on where the coefficient values are located in the quantization matrix.
    Type: Grant
    Filed: November 6, 2012
    Date of Patent: April 30, 2019
    Assignee: QUALCOMM Incorporated
    Inventors: Rajan Laxman Joshi, Marta Karczewicz
  • Patent number: 10264285
    Abstract: In an example, a method of coding video data includes determining, for a pixel associated with a palette index that relates a value of the pixel to a color value in a palette of colors used for coding the pixel, a run length of a run of palette indices being coded with the palette index of the pixel, the method also includes determining a maximum run length for a maximum run of palette indices able to be coded with the palette index of the pixel, and coding data that indicates the run length based on the determined maximum run length.
    Type: Grant
    Filed: May 21, 2015
    Date of Patent: April 16, 2019
    Assignee: Qualcomm Incorporated
    Inventors: Rajan Laxman Joshi, Vadim Seregin, Wei Pu, Marta Karczewicz, Joel Sole Rojals, Krishnakanth Rapaka
  • Patent number: 10244255
    Abstract: A method and apparatus for coding video information are disclosed. In one example, the method may involve: identifying one or more prediction modes for a fixed-rate codec, each prediction mode having a rate-distortion cost; and selecting, for a current block, a prediction mode that has the lowest rate-distortion cost from among the one more prediction modes not previously selected. The method may involve: (i) determining whether the selected prediction mode would result in overflow or underflow of a buffer of the fixed-rate codec; and (ii) determining whether the selected prediction mode would result in remaining bits in the slice being insufficient to code remaining blocks of the slice. The method may involve utilizing a fallback mode to code the current block in response to each of the selected one or more prediction modes resulting in at least one of conditions (i) and (ii) being true.
    Type: Grant
    Filed: March 23, 2016
    Date of Patent: March 26, 2019
    Assignee: QUALCOMM Incorporated
    Inventors: Natan Haim Jacobson, Vijayaraghavan Thirumalai, Rajan Laxman Joshi
  • Patent number: 10212434
    Abstract: A device for coding video data includes a memory storing video data and a video coder including one or more processors configured to determine a current coding unit of the video data is coded in a palette mode; determine a palette for the coding unit by, for a first entry of the palette, choosing a predictor sample from a reconstructed neighboring block of the coding unit and coding a difference between one or more color values of the first entry and one or more color values of the predictor sample.
    Type: Grant
    Filed: January 28, 2016
    Date of Patent: February 19, 2019
    Assignee: Qualcomm Incorporated
    Inventors: Krishnakanth Rapaka, Rajan Laxman Joshi, Vadim Seregin, Marta Karczewicz
  • Patent number: 10200713
    Abstract: An example method of encoding video data includes determining a resolution that will be used for a motion vector that identifies a predictor block in a current picture of video data for a current block in the current picture of video data; determining, based on the determined resolution, a search region for the current block such that a size of the search region is smaller where the resolution is fractional-pixel than where the resolution is integer-pixel; selecting, from within the search region, a predictor block for the current block; determining the motion vector that identifies the selected predictor block for the current block; and encoding, in a coded video bitstream, a representation of the motion vector.
    Type: Grant
    Filed: May 10, 2016
    Date of Patent: February 5, 2019
    Assignee: QUALCOMM Incorporated
    Inventors: Chao Pang, Krishnakanth Rapaka, Vadim Seregin, Marta Karczewicz, Rajan Laxman Joshi
  • Patent number: 10178395
    Abstract: A method of decoding video data, the method comprising receiving a run-length sequence indicative of a binary vector, the binary vector comprising indications of locations of escape samples in a block of video data encoded using a palette-based coding mode, decoding the run-length sequence to obtain the binary vector, and decoding the block of video data using the binary vector. The method of claim 1 may further comprise receiving palette entries for the block of video data receiving one or more escape samples, and receiving a plurality of flags indicating the palette mode for each respective pixel in the block of video data.
    Type: Grant
    Filed: September 29, 2015
    Date of Patent: January 8, 2019
    Assignee: Qualcomm Incorporated
    Inventors: Rajan Laxman Joshi, Vadim Seregin, Wei Pu, Marta Karczewicz, Feng Zou
  • Patent number: 10158836
    Abstract: A device for encoding or decoding video data may clip first residual data based on a bit depth of the first residual data. The device may generate second residual data at least in part by applying an inverse Adaptive Color Transform (IACT) to the first residual data. Furthermore, the device may reconstruct, based on the second residual data, a coding block of a coding unit (CU) of the video data.
    Type: Grant
    Filed: January 26, 2016
    Date of Patent: December 18, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Cheng-Teh Hsieh, Vadim Seregin, Jianle Chen, Rajan Laxman Joshi, Krishnakanth Rapaka, Marta Karczewicz
  • Patent number: 10158866
    Abstract: In an example, a method of processing video data includes determining an input parameter for a truncated binary code that represents a palette index for a pixel of a block of video data based on a number of indices in a palette for the block. The method also includes coding a prefix of the truncated binary code, and determining a palette coding mode for the pixel from a first palette coding mode and a second palette coding mode based only on the prefix of the truncated binary code.
    Type: Grant
    Filed: September 24, 2015
    Date of Patent: December 18, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Wei Pu, Joel Sole Rojals, Marta Karczewicz, Rajan Laxman Joshi, Cheng-Teh Hsieh, Vadim Seregin
  • Publication number: 20180343471
    Abstract: A method and system for encoding video data in a constant bit rate video encoder. A video encoder may determine a coding mode for encoding a block of video data from among one or more coding modes based on a minimum syntax element size and a maximum syntax element size, wherein the minimum syntax element size is greater than 1 bit, encode the block of video data in a plurality of substreams according to the determined coding mode to create a plurality of encoded substreams of video data, store the plurality of encoded substreams of video data in respective balance FIFO buffers, and multiplex the plurality of encoded substreams in a bitstream for transmitting to a decoder.
    Type: Application
    Filed: May 22, 2018
    Publication date: November 29, 2018
    Inventors: Natan Haim Jacobson, Vijayaraghavan Thirumalai, Rajan Laxman Joshi
  • Patent number: 10136141
    Abstract: Techniques are described for palette-based video coding. In palette-based coding, a video coder may form a so-called “palette” as a table of colors for representing video data of a given block of video data. Rather than coding actual pixel values or their residuals for the given block, the video coder may code index values for one or more of the pixels. The index values map the pixels to entries in the palette representing the colors of the pixels. Techniques are described for determining the application of deblocking filtering for pixels of palette coded blocks at a video encoder or a video decoder. In addition, techniques are described for determining quantization parameter (QP) values and delta QP values used to quantize escape pixel values of palette coded blocks at the video encoder or the video decoder.
    Type: Grant
    Filed: June 10, 2015
    Date of Patent: November 20, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Wei Pu, Joel Sole Rojals, Rajan Laxman Joshi, Marta Karczewicz
  • Patent number: 10123045
    Abstract: Methods and apparatuses for modification to block size for transform mode in display stream compression are disclosed. In one aspect, the method involves selecting a current block and a next block of video data, calculating a first rate-distortion (RD) cost for coding the current and next blocks with a larger block size transform (LBT) mode, and calculating a second RD cost for coding each of the current and next blocks with another coding mode different from the LBT mode. The method may further involve determining whether the first RD cost is less than a sum of the second RD costs for the current and next blocks and coding the current and next blocks with the LBT mode in response to the first RD cost being less than the sum of the second RD costs for the current and next blocks.
    Type: Grant
    Filed: July 19, 2016
    Date of Patent: November 6, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Vijayaraghavan Thirumalai, Natan Haim Jacobson, Rajan Laxman Joshi
  • Publication number: 20180302651
    Abstract: A method of content compression including receiving a first block of samples including at least a first sample and a second sample, calculating a predictor value for the first block of samples, calculating a residual between the predictor value and the first sample, quantizing the residual to generate a quantized residual, de-quantizing the quantized residual to generate a de-quantized residual, reconstructing the first sample using the de-quantized residual and the predictor value to generate a first reconstructed sample, calculating an error value based on the first sample and the first reconstructed sample, and modifying the second sample by the error value.
    Type: Application
    Filed: April 11, 2018
    Publication date: October 18, 2018
    Inventors: Natan Haim Jacobson, Ike Ikizyan, Vijayaraghavan Thirumalai, Rajan Laxman Joshi
  • Patent number: 10097837
    Abstract: A video decoder may obtain a syntax element from a bitstream comprising an encoded representation of a block of video data. The block comprises a series of runs, each of which may consist of one pixel or two or more consecutive pixels. Furthermore, the video decoder may obtain a set of palette run length syntax elements indicating lengths of one or more of the runs. The video decoder may determine, based on the syntax element, that the palette run length syntax elements do not include a palette run length syntax element indicating a length of the first run. Rather, the video decoder determines, based on the lengths indicated by the set of palette run syntax elements, the length of the first run. The video decoder may determine, based on the palette sample modes of the pixels, palette indices of the pixels, each indicating an entry in a palette.
    Type: Grant
    Filed: October 8, 2015
    Date of Patent: October 9, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Wei Pu, Feng Zou, Rajan Laxman Joshi, Vadim Seregin, Marta Karczewicz
  • Patent number: 10097842
    Abstract: An apparatus configured to decode video data that comprises a memory configured to store the video data; and one or more processors. The one or more processors are configured to: receive a bitstream that is constrained such that escape values used in a palette mode for decoding the video data are no greater than a maximum value, wherein the bitstream includes information for determining at least one escape value representing a sample in a block of the video data to be decoded; and reconstruct the sample in the block of the video data using the information for determining the at least one escape value.
    Type: Grant
    Filed: September 15, 2016
    Date of Patent: October 9, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Vadim Seregin, Rajan Laxman Joshi, Krishnakanth Rapaka, Marta Karczewicz, Cheng-Teh Hsieh
  • Patent number: 10097839
    Abstract: Techniques are described to extend palette-mode coding techniques to cases where chroma components are at a different resolution than luma components. The entries of the palette table includes three color values and the three color values or a single one of the three color values are selected based on whether a pixel includes both a luma component and chroma components or only a luma component.
    Type: Grant
    Filed: December 17, 2015
    Date of Patent: October 9, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Rajan Laxman Joshi, Vadim Seregin, Wei Pu, Joel Sole Rojals, Marta Karczewicz, Feng Zou
  • Publication number: 20180278951
    Abstract: Techniques are described to increase the coding efficiency of information indicating a sign of a motion vector difference (MVD) component. A video coder may determine a MVD component sign predictor, and determine whether the MVD component sign predictor is the MVD component sign based on an indicator that is transmitted or received. In some examples, the indicator may be context coded, which promotes coding efficiencies.
    Type: Application
    Filed: March 22, 2018
    Publication date: September 27, 2018
    Inventors: Vadim Seregin, Yu-Chen Sun, Yi-Wen Chen, Wei-Jung Chien, Xin Zhao, Rajan Laxman Joshi, Marta Karczewicz
  • Patent number: 10057578
    Abstract: A device for decoding video data is configured to determine, based on a chroma sampling format for the video data, that adaptive color transform is enabled for one or more blocks of the video data; determine a quantization parameter for the one or more blocks based on determining that the adaptive color transform is enabled; and dequantize transform coefficients based on the determined quantization parameter. A device for decoding video data is configured to determine for one or more blocks of the video data that adaptive color transform is enabled; receive in a picture parameter set, one or more offset values in response to adaptive color transform being enabled; determine a quantization parameter for a first color component of a first color space based on a first of the one or more offset values; and dequantize transform coefficients based on the quantization parameter.
    Type: Grant
    Filed: October 6, 2015
    Date of Patent: August 21, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Krishnakanth Rapaka, Li Zhang, Rajan Laxman Joshi, Marta Karczewicz
  • Patent number: 10057587
    Abstract: In one example, a device for decoding video data includes a memory configured to store video data and a video decoder configured to determine that a value for a right shift parameter for an escape-mode coded pixel of a palette-mode coded block of the video data is less than zero, based on the value for the right shift parameter being less than zero, set a value for a left shift parameter to a positive value having an absolute value equal to an absolute value of the right shift parameter, and inverse quantize the escape-mode coded pixel using the value of the left shift parameter.
    Type: Grant
    Filed: January 28, 2016
    Date of Patent: August 21, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Feng Zou, Vadim Seregin, Marta Karczewicz, Wei Pu, Rajan Laxman Joshi
  • Patent number: 10038915
    Abstract: In an example, a method of processing video data includes determining a value of a block-level syntax element that indicates, for all samples of a block of video data, whether at least one respective sample of the block is coded based on a color value of the at least one respective sample not being included in a palette of colors for coding the block of video data. The method also includes coding the block of video data based on the value.
    Type: Grant
    Filed: May 21, 2015
    Date of Patent: July 31, 2018
    Assignee: QUALCOMM Incorporated
    Inventors: Rajan Laxman Joshi, Vadim Seregin, Wei Pu, Marta Karczewicz, Joel Sole Rojals, Krishnakanth Rapaka