비디오 코딩에서 예측 잔차 블록들의 재배치

    公开(公告)号:KR20180041774A

    公开(公告)日:2018-04-24

    申请号:KR20187010756

    申请日:2014-04-23

    Applicant: QUALCOMM INC

    CPC classification number: H04N19/88 H04N19/129 H04N19/159 H04N19/176 H04N19/46

    Abstract: 변환이스킵되거나또는바이패스되는, 손실또는무손실코딩모드에서예측잔차블록의잔차데이터를코딩하는기법들이설명된다. 본개시물의기법들은잔차블록의잔차데이터를코딩하기전에잔차블록의잔차데이터를재배치할지, 예컨대, 회전시키거나또는플리핑할지여부를결정하는것을포함한다. 손실및 무손실코딩모드들양쪽에대해, 비디오코딩디바이스는잔차블록의예측모드및 잔차블록의사이즈에기초하여, 잔차블록의잔차데이터를재배치할지여부를결정한다. 일부예들에서, 본기법들은인터-예측모드및 인트라블록복사예측모드에서모든잔차블록들에대한재배치를디스에이블하고, 그리고, 잔차블록들이임계사이즈이하인블록사이즈들을가질때 인트라-예측모드에서잔차블록들을재배치하도록결정한다.

    Unify intra block copy and inter prediction

    公开(公告)号:NZ728527A

    公开(公告)日:2020-01-31

    申请号:NZ72852715

    申请日:2015-08-21

    Applicant: QUALCOMM INC

    Abstract: The present invention is directed to techniques for improving block vector coding for intra Block Copy (BC), and performing intra BC as part of a unified inter-prediction process for video encoding and video decoding. The techniques described herein may be applied for screen content coding (SCC) or other content where one or more traditional coding tools are inefficient. There are several problems emerged in attempting to unify inter prediction and intra BC. For example, a video encoder may determine that the use of intra BC results in the best rate-distortion metric for a current block when a temporal motion vector predictor (TMVP), which may be a motion vector or a block vector, is used as the block vector of the current block. Hence, the video encoder may signal a candidate index identifying the TMVP and may signal a reference index indicating a position of the current picture in a reference picture list. The video encoder may also signal a motion vector difference (MVD) indicating a difference, if any, between the block vector of the current block and the TMVP (i.e., a motion vector or block vector of the co-located block). However, if the TMVP happens to be a motion vector of sub-pixel accuracy, the motion vector may not be used for intra BC or is inefficient for the use of intra BC because intra BC has integer precision block vectors. The difference in accuracy (i.e., sub-pixel precision versus integer pixel precision) between inter prediction and intra BC also means that these two modes of prediction operate differently because inter prediction operates using sub-pixel precision and intra BC operates using integer pixel precision. The result is an incompatibility undermining the rationale for unifying inter prediction and intra BC. The present invention describes various techniques for implementing a unified intra-block copy and inter-prediction scheme. A video coder may include a current picture and a reference picture in a reference picture list. The video coder may determine a co-located block of the reference picture. The co-located block is co-located with a current block of the current picture. Furthermore, the video coder derives a temporal motion vector predictor from the co-located block and may determine the temporal motion vector predictor has sub-pixel precision. The video coder may right-shift the temporal motion vector predictor determined to have sub-pixel precision. In addition, the video coder may determine, based on the right-shifted temporal motion vector predictor, a predictive block within the current picture. Thus, by converting the TMVP from sub-pixel precision to integer precision, the video coder may be able to use the TMVP as a block vector for use in intra BC to facilitate unification of inter prediction and intra BC.

    Unify intra block copy and inter prediction

    公开(公告)号:AU2015305312B2

    公开(公告)日:2019-11-28

    申请号:AU2015305312

    申请日:2015-08-21

    Applicant: QUALCOMM INC

    Abstract: A video coder may include a current picture and a reference picture in a reference picture list. The video coder may determine a co-located block of the reference picture. The co-located block is co-located with a current block of the current picture. Furthermore, the video coder derives a temporal motion vector predictor from the co-located block and may determine the temporal motion vector predictor has sub-pixel precision. The video coder may right-shift the temporal motion vector predictor determined to have sub-pixel precision. In addition, the video coder may determine, based on the right-shifted temporal motion vector predictor, a predictive block within the current picture.

    Intra BC and inter unification
    5.
    发明专利

    公开(公告)号:AU2015328124B2

    公开(公告)日:2019-04-18

    申请号:AU2015328124

    申请日:2015-10-07

    Applicant: QUALCOMM INC

    Abstract: In general, the disclosure describes techniques related to block vector coding for Intra Block Copy and Inter modes. In one example, the disclosure is directed to a video coding device comprising a memory configured to store video data and one or more processors. The video coding device is configured to determine a reference picture used for coding the current video block and determine a picture order count (POC) value for the reference picture. In response to the POC value for the reference picture being equal to a POC value for a current picture that includes the current video block, the video coding device sets a value of a syntax element to indicate that a reference picture list includes the current picture. Otherwise, the video coding device sets the value of the syntax element to indicate that the reference picture list does not include the current picture.

    previsão de modelo linear com unidades de previsão não-quadradas em código de vídeo

    公开(公告)号:BR112018010664A8

    公开(公告)日:2019-02-26

    申请号:BR112018010664

    申请日:2016-11-23

    Applicant: QUALCOMM INC

    Abstract: um codificador de vídeo reconstrói um conjunto de amostras de referência de croma e reconstrói amostras de luma de uma unidade de previsão não quadrada. adicionalmente, o codificador de vídeo produz sub-amostras do conjunto de amostras de referência de luma de tal modo que um número total das amostras de referência de luma que são vizinhas a um lado mais longo do bloco de previsão não quadrado que é o mesmo que um número total das amostras de referência de luma que são vizinhas a um lado mais curto do bloco de previsão não quadrado. o codificador de vídeo determina um parâmetro de modelo linear (lm) com base na fórmula i, onde i é um número total de amostras de referência no conjunto de amostras de referência de luma, xi é uma amostra de referência de luma no conjunto de referência de luma, yi é uma amostra de referência de croma no conjunto de amostras de referência de croma. o codificador de vídeo utiliza o parâmetro lm em um processo para determinar valores de amostras de croma do bloco de previsão não quadrado.

Patent Agency Ranking