发明名称 VIDEO ENCODING APPARATUS AND VIDEO ENCODING METHOD
摘要 A video encoding apparatus determines whether to substitute all quantized coefficients in a target coefficient group by zero, on the assumption that the subsequent coefficient groups include one or more quantized coefficients that are not zero, the target coefficient group being sequentially selected from the lowest-frequency among coefficient groups included in a block serving as a unit in orthogonal transform, and determines, for each coefficient group sequentially from the lowest-frequency coefficient group, whether to update, to a first quantized-coefficient candidate that is not zero and corresponding to the highest frequency in the coefficient group on the assumption that all the quantized coefficients in the subsequent coefficient groups are zero, a second quantized-coefficient candidate, which is already obtained, based on comparison between the first candidate and the second candidate in terms of a comparison cost obtained by subtracting a coding error related to the subsequent coefficient groups from a cost.
申请公布号 US2016212427(A1) 申请公布日期 2016.07.21
申请号 US201615002351 申请日期 2016.01.20
申请人 FUJITSU LIMITED 发明人 YONEOKA NOBORU
分类号 H04N19/124;H04N19/593;H04N19/65;H04N19/61;H04N19/176 主分类号 H04N19/124
代理机构 代理人
主权项 1. A video encoding apparatus comprising: a processor configured to: divide a plurality of orthogonal transform coefficients included in each of a plurality of blocks into a plurality of coefficient groups each of which includes a predetermined number of the orthogonal transform coefficients, the plurality of blocks being obtained by dividing a picture included in a video, the plurality of orthogonal transform coefficients being obtained by orthogonally transforming, for each block, a prediction error signal obtained on the basis of difference between a value of each pixel of the picture and a prediction signal of the pixel;determine, for each of the predetermined number of orthogonal transform coefficients included in a target coefficient group from among the plurality of coefficient groups, a candidate possible to minimize a cost obtained on the basis of a coding error and an amount of coding among a plurality of quantized-coefficient candidates to be used for quantizing the orthogonal transform coefficient, to be a quantized coefficient of the orthogonal transform coefficient, the target coefficient group being selected from among the plurality of coefficient groups sequentially from the coefficient group including the orthogonal transform coefficients corresponding to lowest frequencies;determine, for a target coefficient group, whether to substitute all the predetermined number of quantized coefficients included in the target coefficient group by zero, on the assumption that a quantized coefficient that is not zero is included in the coefficient group corresponding to higher frequencies than those of the target coefficient group, the target coefficient group being selected from among the plurality of coefficient groups sequentially from the coefficient group including the orthogonal transform coefficients corresponding to the lowest frequencies;determine, for a target coefficient group, a first candidate for the quantized coefficient corresponding to a highest frequency among the quantized coefficients that are included in the target coefficient group and are not zero, on the assumption that all the quantized coefficients included in the coefficient groups corresponding to higher frequencies than those of the target coefficient group are zero, the target coefficient group being selected from among the plurality of coefficient groups sequentially from the coefficient group including the orthogonal transform coefficients corresponding to the lowest frequencies;calculate the coding error of the coefficient groups from the coefficient group adjacent to the coefficient group including a second candidate for the quantized coefficient on a higher frequency side to the coefficient group including the first candidate, the second candidate being the quantized coefficient with the highest frequency among the quantized coefficients not being zero, obtained for the coefficient group corresponding to lower frequencies than those of the coefficient group including the first candidate;update the second candidate to the first candidate when a comparison cost obtained by subtracting the coding error of the coefficient group corresponding to higher frequencies than those of the coefficient group including the first candidate from the cost obtained for the first candidate is lower than a value obtained by adding the coding error to the comparison cost calculated for the second candidate, and determine the second candidate at time when the second candidate for the coefficient group corresponding to highest frequencies among the plurality of coefficient groups is updated, to be the quantized coefficient that is not zero and corresponds to a highest frequency; andcalculate the coding error of the coefficient groups from the coefficient group adjacent to the coefficient group including the second candidate on a higher frequency side to the coefficient group including the first candidate.
地址 KAWASAKI-SHI JP
您可能感兴趣的专利