公开(公告)号：US20240364925A1

公开(公告)日：2024-10-31

申请号：US18636126

申请日：2024-04-15

申请人： QUALCOMM Incorporated

发明人： Hoang Cong Minh LE ,  Qiqi HOU ,  Farzad FARHADZADEH ,  Amir SAID ,  Auke Joris WIGGERS ,  Guillaume Konrad SAUTIERE ,  Reza POURREZA

IPC分类号： H04N19/597 ,  H04N19/137 ,  H04N19/436

CPC分类号： H04N19/597 ,  H04N19/137 ,  H04N19/436

摘要： Systems and techniques are described herein for processing video data. For example, a machine-learning based stereo video coding system can obtain video data including at least a right-view image of a right view of a scene and a left-view image of a left view of the scene. The machine-learning based stereo video coding system can compress the right-view image and the left-view image in parallel to generate a latent representation of the right-view image and the left-view image. The right-view image and the left-view image can be compressed in parallel based on inter-view information between the right-view image and the left-view image, determined using one or more parallel autoencoders.

公开(公告)号：US20240364894A1

公开(公告)日：2024-10-31

申请号：US18604341

申请日：2024-03-13

申请人： DOLBY LABORATORIES LICENSING CORPORATION

发明人： Michael Horowitz

IPC分类号： H04N19/139 ,  C12N9/02 ,  C12P7/14 ,  C12P17/14 ,  H04N19/103 ,  H04N19/105 ,  H04N19/117 ,  H04N19/137 ,  H04N19/159 ,  H04N19/172 ,  H04N19/174 ,  H04N19/196 ,  H04N19/436 ,  H04N19/44 ,  H04N19/46 ,  H04N19/50 ,  H04N19/61 ,  H04N19/70 ,  H04N19/80 ,  H04N19/82 ,  H04N19/91

CPC分类号： H04N19/139 ,  C12N9/0071 ,  C12P7/14 ,  C12P17/14 ,  C12Y114/00 ,  H04N19/103 ,  H04N19/105 ,  H04N19/117 ,  H04N19/137 ,  H04N19/159 ,  H04N19/172 ,  H04N19/174 ,  H04N19/196 ,  H04N19/436 ,  H04N19/44 ,  H04N19/46 ,  H04N19/50 ,  H04N19/61 ,  H04N19/70 ,  H04N19/80 ,  H04N19/82 ,  H04N19/91

摘要： Described is picture segmentation through columns and slices in video encoding and decoding. A video picture is divided into a plurality of columns, each column covering only a part of the video picture in a horizontal dimension. All coded tree blocks (“CTBs”) belonging to a slice may belong to one or more columns. The columns may be used to break the same or different prediction or in-loop filtering mechanisms of the video coding, and the CTB scan order used for encoding and/or decoding may be local to a column. Column widths may be indicated in a parameter set and/or may be adjusted at the slice level. At the decoder, column width may be parsed from the bitstream, and slice decoding may occur in one or more columns.

公开(公告)号：US12132919B2

公开(公告)日：2024-10-29

申请号：US17987844

申请日：2022-11-15

申请人： QUALCOMM Incorporated

发明人： Yang Yang ,  Hoang Cong Minh Le ,  Yinhao Zhu ,  Reza Pourreza ,  Amir Said ,  Yizhe Zhang ,  Taco Sebastiaan Cohen

IPC分类号： H04N19/124 ,  H04N19/119 ,  H04N19/147 ,  H04N19/17 ,  H04N19/436

CPC分类号： H04N19/436 ,  H04N19/119 ,  H04N19/124 ,  H04N19/147 ,  H04N19/17

摘要： A processor-implemented method for image compression using an artificial neural network (ANN) includes receiving, at an encoder of the ANN, an image and a spatial segmentation map corresponding to the image. The spatial segmentation map indicates one or more regions of interest. The encoder compresses the image according to a controllable spatial bit allocation. The controllable spatial bit allocation is based on a learned quantization bin size.

公开(公告)号：US20240357118A1

公开(公告)日：2024-10-24

申请号：US18618551

申请日：2024-03-27

申请人： Alibaba Innovation Private Limited

发明人： Shurun WANG ,  Yan YE

IPC分类号： H04N19/132 ,  H04N19/172 ,  H04N19/186 ,  H04N19/436

CPC分类号： H04N19/132 ,  H04N19/172 ,  H04N19/186 ,  H04N19/436

摘要： A method of encoding a video sequence into a bitstream. The method includes receiving a video sequence; performing a plurality of convolutions on an input image data of the video sequence in YUV format; wherein performing the plurality of convolutions includes performing a first stage convolution on the input image data, wherein the first stage convolution comprises a first convolution and a second convolution that are provided in parallel; performing a second stage convolution on a channel-wise concatenation result of an output of the first convolution and an output of the second convolution; performing a third stage convolution on an output of the second stage convolution; and obtaining an output image data based on an output of the third stage convolution; and encoding the output image data for generating the bitstream.

公开(公告)号：US20240323418A1

公开(公告)日：2024-09-26

申请号：US18238909

申请日：2023-08-28

申请人： Glenfly Tech Co., Ltd.

发明人： Chuanchuan ZHU ,  Shilin YAN ,  Jin SHAO ,  Cong JI

IPC分类号： H04N19/436 ,  H04N19/105 ,  H04N19/176

CPC分类号： H04N19/436 ,  H04N19/105 ,  H04N19/176

摘要： Disclosed are parallel encoding and decoding method and apparatus, a computer device, a storage medium, and a computer program product. The method includes: acquiring a synchronization node corresponding to a codec core, and transmitting a frame synchronization detection request to the synchronization node based on a frame to be encoded and decoded; acquiring a frame synchronization result corresponding to the frame synchronization detection request; and determining a encoding and decoding mode for the frame to be encoded and decoded based on the frame synchronization result, and encoding and decoding the frame to be encoded and decoded based on the encoding and decoding mode. By using the method, a plurality of codec cores can be employed to code and decode different frames in the same video in parallel, thereby improving the frame rate.

公开(公告)号：US20240314363A1

公开(公告)日：2024-09-19

申请号：US18676745

申请日：2024-05-29

申请人： Texas Instruments Incorporated

发明人： Mangesh Devidas Sadafale ,  Minhua Zhou

IPC分类号： H04N19/86 ,  H04N19/117 ,  H04N19/157 ,  H04N19/176 ,  H04N19/436

CPC分类号： H04N19/86 ,  H04N19/117 ,  H04N19/157 ,  H04N19/176 ,  H04N19/436

摘要： Deblocking filtering is provided in which an 8×8 filtering block covering eight sample vertical and horizontal boundary segments is divided into filtering sub-blocks that can be independently processed. To process the vertical boundary segment, the filtering block is divided into top and bottom 8×4 filtering sub-blocks, each covering a respective top and bottom half of the vertical boundary segment. To process the horizontal boundary segment, the filtering block is divided into left and right 4×8 filtering sub-blocks, each covering a respective left and right half of the horizontal boundary segment. The computation of the deviation d for a boundary segment in a filtering sub-block is performed using only samples from rows or columns in the filtering sub-block. Consequently, the filter on/off decisions and the weak/strong filtering decisions of the deblocking filtering are performed using samples contained within individual filtering blocks, thus allowing full parallel processing of the filtering blocks.

公开(公告)号：US20240305797A1

公开(公告)日：2024-09-12

申请号：US18575010

申请日：2022-08-26

申请人： GUANGDONG OPPO MOBILE TELECOMMUNICATIONS CORP., LTD.

发明人： Yue YU ,  Haoping YU

IPC分类号： H04N19/196 ,  H04N19/119 ,  H04N19/169 ,  H04N19/18 ,  H04N19/182 ,  H04N19/186 ,  H04N19/436

CPC分类号： H04N19/197 ,  H04N19/119 ,  H04N19/18 ,  H04N19/182 ,  H04N19/186 ,  H04N19/1883 ,  H04N19/436

摘要： In some embodiments, a video decoder decodes a video from a bitstream of the video using a history-based Rice parameter derivation along with the wavefront parallel processing (WPP). The video decoder accesses a binary string representing a partition of the video and processes each coding tree unit (CTU) in the partition to generate decoded coefficient values in the CTU. The process includes prior to decoding the CTU, determining whether WPP is enabled and the CTU is the first CTU of a current CTU row in the partition, and if so, setting a history counter to an initial value. The process further includes decoding the CTU by calculating the Rice parameters for transform units (TUs) in the CTU based on the value of the history counter and decoding the binary string corresponding to the TUs in the CTU into coefficient values of the TUs based on the calculated Rice parameters.

公开(公告)号：US20240244274A1

公开(公告)日：2024-07-18

申请号：US18620667

申请日：2024-03-28

申请人： HUAWEI TECHNOLOGIES CO., LTD.

发明人： Ahmet Burakhan Koyuncu ,  Atanas Boev ,  Elena Alexandrovna Alshina

IPC分类号： H04N19/91 ,  G06N7/01 ,  H04N19/436

CPC分类号： H04N19/91 ,  G06N7/01 ,  H04N19/436

摘要： Methods and apparatuses are described for entropy encoding and decoding of a latent tensor, which includes separating the latent tensor into segments in the spatial dimensions and in the channel dimension, each segment including at least one latent tensor element. An arrangement of the segments is processed by a neural network; the neural network includes at least one attention layer. Based on the processed segment a probability model is obtained for entropy encoding or decoding of a latent tensor element.

公开(公告)号：US20240236341A1

公开(公告)日：2024-07-11

申请号：US18611971

申请日：2024-03-21

申请人： Panasonic Intellectual Property Corporation of America

发明人： Kiyofumi ABE ,  Takahiro NISHI ,  Tadamasa TOMA

IPC分类号： H04N19/196 ,  H04N19/105 ,  H04N19/14 ,  H04N19/176 ,  H04N19/436

CPC分类号： H04N19/196 ,  H04N19/105 ,  H04N19/14 ,  H04N19/176 ,  H04N19/436

摘要： An encoder includes circuitry and memory connected to the circuitry. In operation, the circuitry: derives a correction parameter using only a neighboring reconstructed image that neighbors a processing unit which has a determined size and is located at an upper left of a current block to be processed in an image, among neighboring reconstructed images that neighbor the current block, and performs correction processing of the current block based on the correction parameter derived, when the current block has a size larger than the determined size.

公开(公告)号：US12028541B2

公开(公告)日：2024-07-02

申请号：US18132032

申请日：2023-04-07

申请人： SHARP KABUSHIKI KAISHA ,  FG Innovation Company Limited

发明人： Sachin G. Deshpande ,  Eiichi Sasaki ,  Takeshi Chujoh ,  Yukinobu Yasugi ,  Tomohiro Ikai ,  Tomoko Aono

IPC分类号： H04N19/436 ,  H04N19/119 ,  H04N19/167 ,  H04N19/174 ,  H04N19/96

CPC分类号： H04N19/436 ,  H04N19/119 ,  H04N19/167 ,  H04N19/174 ,  H04N19/96

摘要： A moving image decoding method is provided for decoding encoded data of a tile group generated by splitting a picture into one or more rectangular regions. The tile group includes one or more segments. The moving image decoding method includes decoding a wavefront parallel processing (WPP) enabled flag indicating whether a rectangular tile or a coding tree unit (CTU) row having a height of one CTU exists in a segment in an object tile group; and decoding an end bit of the segment. When the WPP enabled flag is 1, after decoding a CTU at a right end of the CTU row, an end bit of a first segment having a first fixed value is decoded. When the WPP enabled flag is 0, after decoding a CTU at a bottom right of a tile, an end bit of a second segment having a second fixed value is decoded.