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: 20190273950Abstract: 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×2N 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: ApplicationFiled: December 27, 2018Publication date: September 5, 2019Inventors: Zhao WANG, Siwei MA, Shan GAO, Haitao YANG
-
Patent number: 10298950Abstract: A P frame-based multi-hypothesis motion compensation method includes: taking an encoded image block adjacent to a current image block as a reference image block and obtaining a first motion vector of the current image block by using a motion vector of the reference image block, the first motion vector pointing to a first prediction block; taking the first motion vector as a reference value and performing joint motion estimation on the current image block to obtain a second motion vector of the current image block, the second motion vector pointing to a second prediction block; and performing weighted averaging on the first prediction block and the second prediction block to obtain a final prediction block of the current image block. The method increases the accuracy of the obtained prediction block of the current image block without increasing the code rate.Type: GrantFiled: January 26, 2016Date of Patent: May 21, 2019Assignee: PEKING UNIVERSITY SHENZHEN GRADUATE SCHOOLInventors: Ronggang Wang, Lei Chen, Zhenyu Wang, Siwei Ma, Wen Gao, Tiejun Huang, Wenmin Wang, Shengfu Dong
-
Patent number: 10116934Abstract: An image processing method includes acquiring N pieces of motion information from N adjacent image blocks adjacent to a current image block, where the N adjacent image blocks are in a one-to-one correspondence with the N pieces of motion information, and the N pieces of motion information are in a one-to-one correspondence with the N reference image blocks, determining candidate motion information from the N pieces of motion information according to a preset rule, determining, in the reference image, a location range of a to-be-stored pixel, and storing all pixels in the location range, where the location range covers all pixels of a candidate reference image block, and reading the pixels in the location range, and performing encoding processing on the current image block according to the pixels in the location range.Type: GrantFiled: June 9, 2016Date of Patent: October 30, 2018Assignee: HUAWEI TECHNOLOGIES CO., LTD.Inventors: Jinwen Zan, Sixin Lin, Siwei Ma
-
Publication number: 20180249177Abstract: A reference frame decoding method includes obtaining a first reference frame, where the first reference frame is a first picture frame on which decoding reconstruction has been performed or a first interpolated picture frame obtained by pixel interpolation on the first picture frame, parsing a bitstream to obtain mapping parameters, determining to-be-determined coefficients of a preset mapping function according to the mapping parameters, obtaining, in the first reference frame according to the mapping function whose to-be-determined coefficients have been determined, a first pixel unit having a mapping relationship with a second pixel unit of a second reference frame, and assigning a pixel value of the first pixel unit to the second pixel unit.Type: ApplicationFiled: April 27, 2018Publication date: August 30, 2018Inventors: Sixin Lin, Xiang Zhang, Siwei Ma, Haitao Yang
-
Patent number: 9986255Abstract: A method for video encoding based on an image super-resolution, the method including: 1) performing super-resolution interpolation on a video image to be encoded using a pre-trained texture dictionary database to yield a reference image; in which the texture dictionary database includes: one or multiple dictionary bases, and each dictionary basis includes a mapping group formed by a relatively high resolution image block of a training image and a relatively low resolution image block corresponding to the relatively high resolution image block; 2) performing motion estimation and motion compensation of image blocks of the video image on the reference image to acquire prediction blocks corresponding to the image blocks of the video image; 3) performing subtraction between the image blocks of the video image and the corresponding prediction blocks to yield prediction residual blocks, respectively; and 4) encoding the prediction residual blocks.Type: GrantFiled: March 4, 2016Date of Patent: May 29, 2018Assignee: PEKING UNIVERSITY SHENZHEN GRADUATE SCHOOLInventors: Ronggang Wang, Yang Zhao, Zhenyu Wang, Wen Gao, Wenmin Wang, Shengfu Dong, Tiejun Huang, Siwei Ma
-
Publication number: 20180131960Abstract: Embodiments of the present disclosure disclose a video coding method which includes: obtaining a current picture, motion vector precision of the current picture according to the current picture, where the motion vector precision includes a search step size set, and each search step size in the search step size set is less than a pixel unit; obtaining a current block and a reference picture of the current block; determining a search start point in the reference picture, and searching, from the search start point, for a target integer pixel by using a pixel unit as a search step size; searching, from the target integer pixel, for a match block of the current block according to the motion vector precision; obtaining motion vector information and a prediction residual signal; and the reference picture, the motion vector information, and the prediction residual signal into a bitstream.Type: ApplicationFiled: January 2, 2018Publication date: May 10, 2018Inventors: Zhao Wang, Siwei Ma, Zhuoyi Lv, Haitao Yang
-
Publication number: 20170223351Abstract: A video processing method includes: receiving video data, where the video data is divided into multiple frames; calculating a Lagrange multiplier of a current predict unit in a current coding unit, where the current predict unit is a segment of video signal within the current coding unit, the current coding unit is located in a current frame, and the current frame is one of the multiple frames; performing, by using the Lagrange multiplier of the current predict unit, encoding processing on the current predict unit according to a rate-distortion optimization algorithm to obtain an encoding result of the current predict unit; and sending the encoding result of the current predict unit to a decoder side. An encoding device and a decoding device respectively corresponding to the video processing method are also been provided.Type: ApplicationFiled: April 13, 2017Publication date: August 3, 2017Applicant: HUAWEI TECHNOLOGIES CO., LTD.Inventors: Xiang ZHANG, Siwei MA, Zhuoyi LV
-
Patent number: 9723313Abstract: The embodiments of the present invention provide a method and a device for encoding and decoding videos, and relate to the communication field, and an efficient transformation matrix corresponding to features of each residual block is selected for transformation, which therefore improves encoding efficiency. The solution provided in an embodiment of the present invention is: generating a prediction residual according to input video data; selecting a set of best transformation matrixes among multiple candidate transformation matrixes according to an intra-frame prediction mode and rate-distortion criteria to perform transform-coding on the prediction residual and obtain a transformation result; and generating an encoded stream according to the transformation result and selected transformation matrix index information.Type: GrantFiled: April 20, 2012Date of Patent: August 1, 2017Assignee: HUAWEI TECHNOLOGIES CO., LTD.Inventors: Mingyuan Yang, Dong Wang, Lianhuan Xiong, Xin Zhao, Li Zhang, Siwei Ma, Wen Gao
-
Publication number: 20170200258Abstract: A super-resolution image reconstruction apparatus based on a classified dictionary database. The apparatus can select, from a training image, a first local block and a corresponding second down-sampled local block, extract corresponding features and combine the features into a dictionary group, and perform classification and pre-training on multiple dictionary groups by using calculated values of an LBS and an SES as classification marks, so as to obtain a classified dictionary database of multiple dictionary groups with classification marks. During image reconstruction, local features of a local block on an image to be reconstructed are extracted, the LBS and SES classification of the local block is matched with the LBS and SES classification of each dictionary in the classified dictionary database, so that matched dictionaries can be rapidly obtained, and lastly, image reconstruction is performed on the image to be reconstructed by using the matched dictionaries.Type: ApplicationFiled: May 28, 2014Publication date: July 13, 2017Applicant: PEKING UNIVERSITY SHENZHEN GRADUATE SCHOOLInventors: Yang ZHAO, Ronggang WANG, Zhenyu WANG, Wen GAO, Wenmin WANG, Shengfu DONG, Tiejun HUANG, Siwei MA
-
Publication number: 20170193635Abstract: A method and apparatus for rapidly reconstructing a super-resolution image. In the method and apparatus for rapidly reconstructing a super-resolution image provided in the present application, an original image is processed at least by means of iterative backward mapping based on a texture structural constraint during reconstruction of a super-resolution image of the original image, so as to enhance texture details of the image, thereby improving the high-frequency detail quality of the super-resolution image.Type: ApplicationFiled: May 28, 2014Publication date: July 6, 2017Applicant: PEKING UNIVERSITY SHENZHEN GRADUATE SCHOOLInventors: Yang ZHAO, Ronggang WANG, Zhenyu WANG, Wen GAO, Wenmin WANG, Shengfu DONG, Tiejun HUANG, Siwei MA
-
Patent number: 9485505Abstract: A method and apparatus that determine one or more reference pictures for the current image unit and indicate the reference pictures using information associated with COIs (coding order indexes) of the reference pictures are disclosed. The image unit corresponds to a picture, a slice of the picture, or a region of the picture. Inter-picture encoding or decoding is applied to the input data using the reference pictures. The information associated with the COIs can be incorporated in a slice header or a picture header of a bitstream associated with the video sequence. Furthermore, the COI can be represented by a coded COI using a fixed number of bits, wherein the coded COI is constrained to a range from 0 to MAC_COI-1 and MAX_COI is a positive integer.Type: GrantFiled: December 12, 2013Date of Patent: November 1, 2016Assignee: MEDIATEK SINGAPORE PTE. LTD.Inventors: Kai Zhang, Jicheng An, Qin Yu, Siwei Ma
-
Publication number: 20160295213Abstract: An image processing method includes acquiring N pieces of motion information from N adjacent image blocks adjacent to a current image block, where the N adjacent image blocks are in a one-to-one correspondence with the N pieces of motion information, and the N pieces of motion information are in a one-to-one correspondence with the N reference image blocks, determining candidate motion information from the N pieces of motion information according to a preset rule, determining, in the reference image, a location range of a to-be-stored pixel, and storing all pixels in the location range, where the location range covers all pixels of a candidate reference image block, and reading the pixels in the location range, and performing encoding processing on the current image block according to the pixels in the location range.Type: ApplicationFiled: June 9, 2016Publication date: October 6, 2016Inventors: Jinwen Zan, Sixin Lin, Siwei Ma
-
Publication number: 20160212448Abstract: A method for video encoding based on a dictionary database, the method including: 1) dividing a current image frame to be encoded in a video stream into a plurality of image blocks; 2) recovering encoding distortion information of a decoded and reconstructed image of a previous frame of the current image frame using a texture dictionary database to obtain an image with recovered encoding distortion information, and performing temporal prediction using the image with the recovered encoding distortion information as a reference image to obtain prediction blocks of image blocks to be encoded; in which, the texture dictionary database includes: clear image dictionaries and distorted image dictionaries corresponding to the clear image dictionaries; and 3) performing subtraction between the image blocks to be encoded and the prediction blocks to obtain residual blocks, and processing the residual blocks to obtain a video bit stream.Type: ApplicationFiled: March 27, 2016Publication date: July 21, 2016Inventors: Ronggang WANG, Yang ZHAO, Zhenyu WANG, Wen GAO, Wenmin WANG, Shengfu DONG, Tiejun HUANG, Siwei MA
-
Publication number: 20160191940Abstract: A method for video encoding based on an image super-resolution, the method including: 1) performing super-resolution interpolation on a video image to be encoded using a pre-trained texture dictionary database to yield a reference image; in which the texture dictionary database includes: one or multiple dictionary bases, and each dictionary basis includes a mapping group formed by a relatively high resolution image block of a training image and a relatively low resolution image block corresponding to the relatively high resolution image block; 2) performing motion estimation and motion compensation of image blocks of the video image on the reference image to acquire prediction blocks corresponding to the image blocks of the video image; 3) performing subtraction between the image blocks of the video image and the corresponding prediction blocks to yield prediction residual blocks, respectively; and 4) encoding the prediction residual blocks.Type: ApplicationFiled: March 4, 2016Publication date: June 30, 2016Inventors: Ronggang WANG, Yang ZHAO, Zhenyu WANG, Wen GAO, Wenmin WANG, Shengfu DONG, Tiejun HUANG, Siwei MA
-
Publication number: 20160142729Abstract: A coding method based on multi-hypothesis motion compensation for a P-frame, including: a) using neighboring coded image blocks as reference image blocks, adopting a motion vector of each reference image block as a first motion vector which points to a first prediction block; b) adopting the first prediction block corresponding to each reference image block as a reference value, and performing joint motion estimation on the current image block to acquire a second motion vector which points to a second prediction block; c) weighted averaging the first prediction block and the second prediction corresponding to each reference image block to acquire a third prediction block of the current image block, respectively; and d) calculating a coding cost corresponding to each reference image block to determine a final first motion vector, a final second motion vector, and a final prediction block of the current image block.Type: ApplicationFiled: January 26, 2016Publication date: May 19, 2016Inventors: Ronggang WANG, Lei CHEN, Zhenyu WANG, Siwei MA, Wen GAO, Tiejun HUANG, Wenmin WANG, Shengfu DONG
-
Publication number: 20160142728Abstract: A P frame-based multi-hypothesis motion compensation method includes: taking an encoded image block adjacent to a current image block as a reference image block and obtaining a first motion vector of the current image block by using a motion vector of the reference image block, the first motion vector pointing to a first prediction block; taking the first motion vector as a reference value and performing joint motion estimation on the current image block to obtain a second motion vector of the current image block, the second motion vector pointing to a second prediction block; and performing weighted averaging on the first prediction block and the second prediction block to obtain a final prediction block of the current image block. The method increases the accuracy of the obtained prediction block of the current image block without increasing the code rate.Type: ApplicationFiled: January 26, 2016Publication date: May 19, 2016Inventors: Ronggang Wang, Lei Chen, Zhenyu Wang, Siwei Ma, Wen Gao, Tiejun Huang, Wenmin Wang, Shengfu Dong
-
Patent number: 9271006Abstract: A coding and decoding method for images or videos is provided by embodiments of the present invention to improve coding and decoding efficiency. The method includes: establishing a visual dictionary, wherein, the visual dictionary includes one or more visual words; extracting features from a specific object in an image; determining whether there is a visual word in the visual dictionary matching the specific object by using a feature matching method; obtaining the index of the visual word matched and a geometric relationship between the specific object and the visual word matched, wherein, the geometric relationship is represented by a project parameter; entropy coding the index of the visual word matched and the project parameter instead of entropy coding the specific object.Type: GrantFiled: November 6, 2014Date of Patent: February 23, 2016Assignee: PEKING UNIVERSITYInventors: Tiejun Huang, Wen Gao, Siwei Ma
-
Patent number: 9088797Abstract: A video processing apparatus with residue prediction includes a motion estimation/compensation unit to determine a matching block of a reference video frame, obtain a motion vector of a current block of a current video frame that is related to the matching block, and acquire neighboring reconstructed pixels adjacent to the current block and corresponding pixels adjacent to the matching block with the motion vector alignment. Additionally, a pseudo-residue generating unit is included and constructs pseudo residues according to the neighboring reconstructed pixels and the corresponding pixels, an arithmetic unit is included and generates first-order residues by subtracting the matching block from the current block, and a residue-predicting unit is included and derives second-order residues and corresponding information according to the pseudo residues and the first-order residues.Type: GrantFiled: December 19, 2008Date of Patent: July 21, 2015Assignee: MEDIATEK SINGAPORE PTE. LTD.Inventors: Kai Zhang, Li Zhang, Siwei Ma, Wen Gao, Shaw-Min Lei
-
Publication number: 20150131921Abstract: A coding and decoding method for images or videos is provided by embodiments of the present invention to improve coding and decoding efficiency. The method includes: establishing a visual dictionary, wherein, the visual dictionary includes one or more visual words; extracting features from a specific object in an image; determining whether there is a visual word in the visual dictionary matching the specific object by using a feature matching method; obtaining the index of the visual word matched and a geometric relationship between the specific object and the visual word matched, wherein, the geometric relationship is represented by a project parameter; entropy coding the index of the visual word matched and the project parameter instead of entropy coding the specific object.Type: ApplicationFiled: November 6, 2014Publication date: May 14, 2015Inventors: Tiejun Huang, Wen Gao, Siwei Ma
-
Publication number: 20140169459Abstract: A method and apparatus that determine one or more reference pictures for the current image unit and indicate the reference pictures using information associated with COIs (coding order indexes) of the reference pictures are disclosed. The image unit corresponds to a picture, a slice of the picture, or a region of the picture. Inter-picture encoding or decoding is applied to the input data using the reference pictures. The information associated with the COIs can be incorporated in a slice header or a picture header of a bitstream associated with the video sequence. Furthermore, the COI can be represented by a coded COI using a fixed number of bits, wherein the coded COI is constrained to a range from 0 to MAC_COI-1 and MAX_COI is a positive integer.Type: ApplicationFiled: December 12, 2013Publication date: June 19, 2014Applicant: Media Tek Singapore Pte. Ltd.Inventors: Kai Zhang, Jicheng An, Qin Yu, Siwei Ma