专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明公开

EP4350538A1 FEATURE MAP PROCESSING METHOD AND RELATED DEVICE 审中-实审
公开(公告)号：EP4350538A1
公开(公告)日：2024-04-10
申请号：EP22832005.7
申请日：2022-06-28
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , MAO, Jue , YANG, Haitao
IPC分类号： G06F18/00
摘要： This application relates to the field of artificial intelligence technologies, and provides a feature map processing method and a related device. The method is implemented by invoking a neural network model that includes a plurality of input adaptation branches and a post-processing part. An output of each of the plurality of input adaptation branches is an input of the post-processing part, and each of the plurality of input adaptation branches can downsample a feature map at a different ratio. Specifically, a first feature map is input to one of the plurality of input adaptation branches, to obtain a second feature map, where the one input adaptation branch is determined based on a size of the first feature map and a preset size threshold, the preset size threshold is an upper limit size of an input feature map of the post-processing part, and a size of the second feature map is not greater than the preset size threshold; and the second feature map is input to the post-processing part for post-processing, to obtain a processing result of the first feature map. In embodiments of this application, computing power can be reduced.

2. 发明公开

EP4246971A1 QUANTIZATION PARAMETER DECODING METHOD AND DEVICE 审中-实审
公开(公告)号：EP4246971A1
公开(公告)日：2023-09-20
申请号：EP21900062.7
申请日：2021-12-02
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , CHEN, Shaolin , YANG, Haitao
IPC分类号： H04N19/124
摘要： Embodiments of the present invention provide a quantization parameter decoding method and an apparatus, and relate to the field of video or image compression technologies. The method includes: determining a current rectangular region to which a current CU belongs, where the current rectangular region is one of a plurality of rectangular regions included in a to-be-decoded CTU, and all of the plurality of rectangular regions meet a preset condition; determining a target context model based on a quantity of luminance QP differential values that have been obtained through decoding in the current rectangular region; decoding a bitstream based on the target context model to obtain a target binary symbol, where the target binary symbol indicates whether a luminance QP differential value of the current CU is 0, and the target binary symbol is a binary symbol of the luminance QP differential value of the current CU; determining the luminance QP differential value of the current CU based on the target binary symbol; and obtaining a luminance QP of the current CU based on the luminance QP differential value of the current CU and a luminance QP prediction value of the current CU. Embodiments of the present invention can effectively reduce bit overheads of the luminance QP differential value in the bitstream, and improve coding efficiency.

3. 发明公开

EP3955575A1 INTRA-FRAME PREDICTION METHOD AND APPARATUS FOR VIDEO SEQUENCE 有权
公开(公告)号：EP3955575A1
公开(公告)日：2022-02-16
申请号：EP20799360.1
申请日：2020-03-27
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , YANG, Haitao , CHEN, Jianle
IPC分类号： H04N19/176 , H04N19/159 , H04N19/103 , H04N19/117
摘要： Embodiments of this application disclose an intra prediction method for a video sequence. The method includes: obtaining a size and a split mode of a chroma coding tree node; determining, based on the size and the split mode of the chroma coding tree node, whether to allow cross-component linear model (CCLM) intra prediction to be performed on a chroma coding unit included in the chroma coding tree node; determining a prediction mode of the chroma coding unit depending on whether to allow CCLM to be performed on the chroma coding unit; and processing the chroma coding unit based on the prediction mode of the chroma coding unit to obtain prediction samples of the chroma coding unit. The embodiments of this application further disclose an intra prediction apparatus and a decoder. The embodiments of this application help reduce a latency in processing a chroma coding unit, and improve a throughput of a hardware decoder.

4. 发明公开

EP3886438A1 FLAG BIT CONTEXT MODELING METHOD AND DEVICE 审中-实审
公开(公告)号：EP3886438A1
公开(公告)日：2021-09-29
申请号：EP19902637.8
申请日：2019-12-23
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , ZHANG, Lian , YANG, Haitao , CHEN, Jianle
IPC分类号： H04N19/91
摘要： A context modeling method and apparatus for a flag are provided, and relate to the video coding field. The method includes: determining, for a current block, whether a first preset condition is satisfied and whether a second preset condition is satisfied (S101), where the first preset condition is a preset condition related to a first neighboring block neighboring to the current block in a first direction, the second preset condition is a preset condition related to a second neighboring block neighboring to the current block in a second direction, and the first direction is perpendicular to the second direction; determining a subclass index of a context model of a first flag of the current block based on the height and the width of the current block (S 102); and determining the context model of the first flag of the current block based on a satisfaction status of the first preset condition, a satisfaction status of the second preset condition, and the subclass index of the context model of the first flag (S103). According to the method, accuracy of determining a context model of a flag can be improved to some extent, so that coding performance is improved.

5. 发明公开

EP3844966A1 ENCODING METHOD, DECODING METHOD, ENCODER, AND DECODER 审中-公开
公开(公告)号：EP3844966A1
公开(公告)日：2021-07-07
申请号：EP18935533.2
申请日：2018-09-27
申请人： HUAWEI TECHNOLOGIES CO., LTD.
发明人： GAO, Han , ZHAO, Yin , ESENL K, Semih
IPC分类号： H04N19/55 , H04N19/61

6. 发明公开

EP3840378A1 VIDEO DECODING METHOD AND VIDEO DECODER 审中-公开
公开(公告)号：EP3840378A1
公开(公告)日：2021-06-23
申请号：EP19857345.3
申请日：2019-08-28
申请人： HUAWEI TECHNOLOGIES CO., LTD.
发明人： ZHAO, Yin , YANG, Haitao
IPC分类号： H04N19/124 , H04N19/96 , H04N19/70
摘要： The present invention discloses a video decoding method and a video decoder. The method includes: parsing coding tree split information to obtain a current node; determining, based on a depth N of the current node, a region covered by a current quantization group; obtaining a quantization parameter QP delta of a current coding unit CU in the region covered by the current quantization group; and obtaining a reconstructed picture of the current CU based on the QP delta of the current CU. By using the present invention, decoding efficiency can be improved.

7. 发明公开

EP3679715A1 CODING METHOD AND APPARATUS 有权
公开(公告)号：EP3679715A1
公开(公告)日：2020-07-15
申请号：EP18867653.0
申请日：2018-10-07
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , YANG, Haitao , CHEN, Jianle
IPC分类号： H04N19/44

8. 发明公开

EP3618441A1 METHOD AND DEVICE FOR MOTION VECTOR PREDICTION 审中-公开
公开(公告)号：EP3618441A1
公开(公告)日：2020-03-04
申请号：EP18817683.8
申请日：2018-03-23
申请人： Huawei Technologies Co., Ltd.
发明人： MAO, Jue , YU, Lu , ZHAO, Yin
IPC分类号： H04N19/139 , H04N19/513
摘要： Embodiments of this application disclose a motion vector prediction method and device, to determine a motion vector predictor of a current block, so as to improve motion vector prediction efficiency. The method in the embodiments of this application includes: obtaining a reference frame corresponding to a motion vector of a current block; determining a motion vector type of the motion vector of the current block based on a reference frame type of the reference frame; obtaining a motion vector type of a first motion vector of at least one spatially adjacent prediction block of the current block when the at least one spatially adjacent prediction block is an inter-frame prediction block; and when the obtained motion vector type of the first motion vector is different from all motion vector types of motion vectors of the current block, determining a second motion vector of a spatially adjacent pixel block of the current block based on decoding information of the adjacent pixel block, and using the second motion vector or a scaling value of the second motion vector as a candidate motion vector predictor of the current block.

9. 发明公开

EP3334160A1 METHOD AND APPARATUS FOR DE-QUANTIZATION OF TRANSFORM COEFFICIENTS, AND DECODING DEVICE 审中-公开
公开(公告)号：EP3334160A1
公开(公告)日：2018-06-13
申请号：EP15907652.0
申请日：2015-11-06
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , YANG, Haitao
IPC分类号： H04N19/124 , H04N19/176 , H04N19/157
CPC分类号： H04N19/124 , H04N19/122 , H04N19/157 , H04N19/176 , H04N19/18 , H04N19/44 , H04N19/91
摘要： The present invention discloses a transform coefficient dequantization method. A unit size of a to-be-processed unit corresponding to a first transform coefficient set is determined by using division information. Then, a quantization regulation factor for a first transform coefficient set is determined according to a first preset algorithm and the unit size of the to-be-processed unit, and with the first preset algorithm, the quantization regulation factor decreases progressively with the size of the to-be-processed unit. Dequantization processing is performed on a transform coefficient in the first transform coefficient set according to a second preset algorithm by using the quantization regulation factor. Based on the transform coefficient dequantization method disclosed in the present invention, a decoding device adaptively determines the quantization regulation factor for the first transform coefficient set according to the unit size of the to-be-processed unit, and then performs dequantization processing on the transform coefficient in the first transform coefficient set by using the determined quantization regulation factor. A code stream sent by the coding device does not need to carry quantization step size regulation information, and therefore coding efficiency can be improved. The present invention further discloses a corresponding apparatus and a decoding device.

10. 发明公开

EP3236658A1 IMAGE PROCESSING METHOD AND DEVICE 有权
标题翻译：图像处理方法和装置
公开(公告)号：EP3236658A1
公开(公告)日：2017-10-25
申请号：EP16734909.1
申请日：2016-01-06
申请人： Huawei Technologies Co., Ltd.
发明人： ZHAO, Yin , YANG, Haitao
IPC分类号： H04N19/60
CPC分类号： H04N19/60 , H04N19/136 , H04N19/18 , H04N19/182 , H04N19/50 , H04N19/61 , H04N19/80
摘要： An image processing method and apparatus are disclosed. The method includes: transforming a first image to obtain a first transform coefficient set; transforming a second image, or transforming a difference image between the first image and a second image, to obtain a second transform coefficient set; selecting a first group of transform coefficients from the first transform coefficient set according to magnitudes of transform coefficients in the first transform coefficient set, where magnitudes of transform coefficients in the first group of transform coefficients meet a preset threshold condition; determining an image difference between the first image and the second image according to the first group of transform coefficients and the second transform coefficient set; and processing the first image and the second image according to the image difference. An image difference determined in embodiments can well reflect a difference degree between two images, so that subsequent image processing based on the image difference is more accurate.
摘要翻译：公开了一种图像处理方法和装置。该方法包括：变换第一图像以获得第一变换系数集; 变换第二图像或变换第一图像与第二图像之间的差异图像以获得第二变换系数集合; 根据所述第一变换系数集合中的变换系数的大小，从所述第一变换系数集合中选择第一变换系数组，其中所述第一变换系数组中的变换系数的大小满足预设阈值条件; 根据所述第一组变换系数和所述第二变换系数组确定所述第一图像与所述第二图像之间的图像差异; 以及根据图像差异处理第一图像和第二图像。在实施例中确定的图像差异可以很好地反映两个图像之间的差异程度，使得基于图像差异的后续图像处理更准确。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式