Patents by Inventor Siwei Ma
Siwei Ma 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).
-
Publication number: 20220394308Abstract: A method implemented by a video coding apparatus includes applying a neural network (NN) filter to an unfiltered sample of a video unit to generate a filtered sample. The NN filter is applied based on a syntax element of the video unit. The method also includes converting between a video media file and a bitstream based on the filtered sample that was generated.Type: ApplicationFiled: April 13, 2022Publication date: December 8, 2022Inventors: Yue Li, Li Zhang, Kai Zhang, Junru Li, Meng Wang, Siwei Ma, Shiqi Wang
-
Publication number: 20220345699Abstract: An in-loop filtering method includes determining a target filter for a chrominance component of a current block of an image from a plurality of cross-component ALF filters, determining target filter coefficients for the chrominance component according to the chrominance component after ALF and a luminance component of the current block without ALF, filtering the chrominance component after ALF according to the target filter and the target filter coefficients, determining a filtered chrominance component of the current block according to the chrominance component after being filtered with the target filter coefficients and the chrominance component after the ALF, encoding according to the filtered chrominance component of the current block, and encoding a total number of the plurality of cross-component ALF filters as a syntax element. A stream of the image includes only one syntax element for indicating the total number of the plurality of cross-component ALF filters.Type: ApplicationFiled: June 29, 2022Publication date: October 27, 2022Inventors: Siwei MA, Xuewei MENG, Xiaozhen ZHENG, Shanshe WANG
-
Publication number: 20220321907Abstract: A video image processing method includes, in response to a specific neighboring block of a current image block satisfying a preset condition, determining a collocated block of the current image block according to a motion vector of the specific neighboring block, determining a specific candidate from a motion vector candidate list of the current image block according to the collocated block of the current image block for using in prediction for the current image block, scanning neighboring blocks of the current image block and determining a control point motion vector group of each of one or more neighboring blocks that use an affine transformation mode for prediction, and adding the control point motion vector group of each of the determined one or more neighboring blocks into the motion vector candidate list of the current image block as a candidate.Type: ApplicationFiled: June 20, 2022Publication date: October 6, 2022Inventors: Xiaozhen ZHENG, Tianliang FU, Shanshe WANG, Siwei MA, Weiran LI, Suhong WANG
-
Publication number: 20220286706Abstract: A video processing method includes obtaining motion information of a neighboring block of a current image block, dividing the current image block into a plurality of sub-blocks in response to the neighboring block satisfying a preset condition, determining, in a time-domain reference image of the current image block, related blocks of the plurality of sub-blocks according to a motion vector of the neighboring block, and performing prediction on the current image block according to motion vectors of the related blocks.Type: ApplicationFiled: May 23, 2022Publication date: September 8, 2022Inventors: Xiaozhen ZHENG, Suhong WANG, Shanshe WANG, Siwei MA, Weiran LI
-
Patent number: 11394966Abstract: A video encoding and decoding method includes determining an image block set based on one or more similar image blocks of an image block to be processed, generating an initial structure matrix of the image block set based on a pixel value of each image block in the image block set, and performing a filtering process on the initial structure matrix to obtain a reconstructed structure matrix. The reconstructed structure matrix corresponds to a reconstructed pixel value of the each image block. The filtering process includes one or more point-fixing processes. The method further includes reconstructing the each image block based on the reconstructed pixel value of the each image block.Type: GrantFiled: October 1, 2020Date of Patent: July 19, 2022Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Siwei Ma, Xuewei Meng, Shanshe Wang, Chuanmin Jia, Xiaozhen Zheng
-
Patent number: 11368714Abstract: A video image processing method includes dividing a coding unit into one or more sub-block, constructing first and second motion vector candidate lists, and performing prediction for the coding unit according to the first and second motion vector candidate lists. The first motion vector candidate list includes a motion vector of a sub-block of the coding unit, and the second motion vector candidate list includes a motion vector of the coding unit. Constructing the first motion vector candidate list includes determining a reference motion vector of the coding unit, determining a related reference block of the sub-block of the coding unit in a co-located reference image of the coding unit according to the reference motion vector, determining a scaling factor of the motion vector of the related reference block, adding the scaled motion vector into the first motion vector candidate list.Type: GrantFiled: September 30, 2020Date of Patent: June 21, 2022Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Xiaozhen Zheng, Tianliang Fu, Shanshe Wang, Siwei Ma, Weiran Li, Suhong Wang
-
Patent number: 11343534Abstract: A video processing method includes performing or skipping following processing based on whether a size of a coding unit is not smaller or is smaller than 8×8. The processing includes dividing the coding unit into sub-blocks each having a fixed size of 8×8, scanning a left neighboring block and determining a reference motion vector of the coding unit, determining a related reference block of a sub-block in the co-located reference image according to the reference motion vector, determining a scaling factor of a motion vector of the related reference block, scaling the motion vector of the related reference block using the scaling factor, determining motion information of the sub-block according to the scaled motion vector, and performing prediction for the coding unit according to the motion information.Type: GrantFiled: April 1, 2021Date of Patent: May 24, 2022Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Xiaozhen Zheng, Suhong Wang, Shanshe Wang, Siwei Ma, Weiran Li
-
Patent number: 11323742Abstract: A video processing method includes performing or skipping following processing based on whether a size of a coding unit is not smaller or is smaller than 8×8. The processing includes dividing the coding unit into sub-blocks each having a fixed size of 8×8, scanning a left neighboring block and determining a reference motion vector of the coding unit, determining a related reference block of a sub-block in the co-located reference image according to the reference motion vector, determining a scaling factor of a motion vector of the related reference block, scaling the motion vector of the related reference block using the scaling factor, determining motion information of the sub-block according to the scaled motion vector, and performing prediction for the coding unit according to the motion information.Type: GrantFiled: September 30, 2020Date of Patent: May 3, 2022Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Xiaozhen Zheng, Suhong Wang, Shanshe Wang, Siwei Ma, Weiran Li
-
Publication number: 20220124346Abstract: A video image processing method includes binarizing a chrominance prediction mode of the image block of a to-be-encoded/decoded image to obtain a bit string, the bit string including two adjacent bits, the to-be-encoded/decoded image being permitted to use a CCLM and/or a regular intra-frame chrominance prediction mode, and the CCLM including a first mode, a second mode, and a third mode, and using mutually independent probability models to encode/decode a first bit and a second bit. The first bit of the adjacent bits indicates whether the CCLM is used, the second bit indicates whether to use the first mode of the CCLM when the first bit indicates to use the CCLM, the third bit indicates whether the second mode or the third mode of the CCLM is used when the second bit indicates that the first mode of the CCLM is not used.Type: ApplicationFiled: December 23, 2021Publication date: April 21, 2022Inventors: Siwei MA, Jiaqi ZHANG, Xiaozhen ZHENG, Shanshe WANG
-
Publication number: 20220116644Abstract: A video image processing method including determining a current image block, in response to a size of the current image block meeting a preset condition, skipping a temporal motion vector prediction (TMVP) operation so that a temporal candidate motion vector of the current image block is not determined according to the TMVP operation, and encoding the current image block. The TMVP operation includes determining a relevant block of the current image block in a temporal neighboring image, and determining the temporal candidate motion vector of the current image block according to a motion vector of the relevant block.Type: ApplicationFiled: December 20, 2021Publication date: April 14, 2022Inventors: Xiaozhen ZHENG, Suhong WANG, Shanshe WANG, Siwei MA
-
Publication number: 20220116664Abstract: The present disclosure provides a loop filtering method which includes determining a clip index parameter of a loop filter, and encoding the clip index parameter by using a non-exponential Golomb encoding method.Type: ApplicationFiled: December 23, 2021Publication date: April 14, 2022Inventors: Siwei MA, Xuewei MENG, Xiaozhen ZHENG, Shanshe WANG
-
Publication number: 20220086483Abstract: A video encoding method including obtaining a motion vector candidate list of a current image block; determining a base motion vector according to motion vectors in the motion vector candidate list; in response to the base motion vector pointing to a short-term reference image, scaling a motion vector offset of the base motion vector using a scaling factor to obtain a scaled motion vector offset, and performing prediction for the current image block according to a motion vector determined by the scaled motion vector offset and the base motion vector; and in response to the base motion vector pointing to a long-term reference image, skipping scaling of the motion vector offset, and performing prediction for the current image block according to a motion vector determined by the motion vector offset that is not scaled and the base motion vector.Type: ApplicationFiled: November 29, 2021Publication date: March 17, 2022Inventors: Xiaozhen ZHENG, Suhong WANG, Shanshe WANG, Siwei MA, Weiran LI
-
Publication number: 20220078466Abstract: A video image processing method includes determining a current image block; in response to a size of the current image block not meeting a first preset condition, skipping an advanced/alternative temporal motion vector prediction (ATMVP) operation so that a temporal candidate motion vector of the current image block is not determined according to the ATMVP operation; and encoding the current image block. The ATMVP operation includes determining a relevant block of the current image block in a temporal neighboring image; dividing the current image block into a plurality of sub-blocks; for each of the sub-blocks, determining a corresponding relevant sub-block in the relevant block; and determining temporal candidate motion vectors of the sub-blocks of the current image block according to motion vectors of the relevant sub-blocks corresponding to the sub-blocks.Type: ApplicationFiled: November 15, 2021Publication date: March 10, 2022Inventors: Suhong WANG, Xiaozhen ZHENG, Shanshe WANG, Siwei MA
-
Patent number: 11272204Abstract: A motion compensation method includes determining one or more weighting coefficients of a predicted value of a target pixel to be processed according to at least one of a first parameter or a second parameter, and determining the predicted value of the target pixel according to the weighting coefficient. The target pixel is in a boundary pixel block of a current image block. The first parameter is a size of the current image block or a distance between the target pixel and a center position of the current image block. The second parameter is a size of an adjacent image block of the current image block or a distance between the target pixel and a center position of the adjacent image block.Type: GrantFiled: August 13, 2020Date of Patent: March 8, 2022Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Zhao Wang, Siwei Ma, Xiaozhen Zheng
-
Patent number: 11245932Abstract: A decoding method includes: parsing a data stream, and if partitioning an image block with a size of 2N×2N using a quadtree partition pattern is allowed, processing a 2N×N first subimage block and a 2N×N second subimage block or an N×2N first subimage block and an N×2N second subimage block in a constraint subimage processing mode, wherein an image block partition pattern obtained for the partitioned second subimage block and the partitioned first subimage block is different from an image block partition pattern obtained after the 2N×2N image block is partitioned using the quadtree partition pattern, where the 2N×N first subimage block and the 2N×N second subimage block or the N×2N first subimage block and the N×2N second subimage block are obtained by partitioning the image block with the size of 2N×2N.Type: GrantFiled: October 9, 2020Date of Patent: February 8, 2022Assignee: Huawei Technologies Co., Ltd.Inventors: Zhao Wang, Siwei Ma, Shan Gao, Haitao Yang
-
Publication number: 20210409720Abstract: A video processing method includes constructing a motion information candidate list for an image block of a current frame, selecting motion information for a sub-image-block included in the image block from the motion information candidate list, and encoding or decoding the image block according to the selected motion information. The motion information candidate list includes at least one piece of dual motion-information.Type: ApplicationFiled: September 13, 2021Publication date: December 30, 2021Inventors: Xuewei MENG, Xiaozhen ZHENG, Shanshe WANG, Siwei MA
-
Patent number: 11206422Abstract: A video image processing method includes determining a current image block and, in response to a size of the current image block meeting a preset condition, determining a temporal candidate motion vector of the current image block according to at least one of a temporal motion vector prediction (TMVP) operation or an advanced/alternative temporal motion vector prediction (ATMVP) operation.Type: GrantFiled: September 30, 2020Date of Patent: December 21, 2021Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Xiaozhen Zheng, Suhong Wang, Shanshe Wang, Siwei Ma
-
Patent number: 11190798Abstract: A video image processing method includes dividing a coding tree unit into one or more coding units using a quadtree plus binary tree division method, obtaining a motion vector candidate list of one of the one or more coding units, determining a base motion vector according to motion vectors in the motion vector candidate list, in response to the base motion vector pointing to a short-term reference image, scaling the base motion vector using a scaling factor not equal to 1, in response to the base motion vector pointing to a long-term reference image, scaling the base motion vector using a scaling factor set to 1, and performing prediction for the one of the one or more coding units according to the base motion vector after being scaled.Type: GrantFiled: September 30, 2020Date of Patent: November 30, 2021Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Xiaozhen Zheng, Suhong Wang, Shanshe Wang, Siwei Ma, Weiran Li
-
Patent number: 11178420Abstract: A video image processing method includes determining a current image block and, in response to a size of the current image block meeting a preset condition, determining a temporal candidate motion vector of the current image block according to at least one of a temporal motion vector prediction (TMVP) operation or an advanced/alternative temporal motion vector prediction (ATMVP) operation.Type: GrantFiled: September 30, 2020Date of Patent: November 16, 2021Assignee: SZ DJI TECHNOLOGY CO., LTD.Inventors: Suhong Wang, Xiaozhen Zheng, Shanshe Wang, Siwei Ma
-
Publication number: 20210337232Abstract: A video processing method includes obtaining a motion vector of a spatial neighboring block of a current block as an initial temporal motion vector. The current block is an image block using bidirectional prediction. The method further includes determining a first reference frame list and a second reference frame list of the current block, obtaining a temporal motion vector of the current block, determining a corresponding block of the current block in the reference frame according to the temporal motion vector of the current block, determining motion information of a sub-block of the current block according to the corresponding block of the current block in the reference frame, adding the motion information of the sub-black of the current block into an affine merge candidate list, and performing inter-frame prediction on the current block according to the affine merge candidate list.Type: ApplicationFiled: July 1, 2021Publication date: October 28, 2021Inventors: Xiaozhen ZHENG, Suhong WANG, Siwei MA, Shanshe WANG