基于时域相关性的自适应采样值补偿算法

Sample Adaptive Offset Algorithm Based on Temporal Correlation

虽然自适应采样值补偿(SAO)滤波器能有效提高视频的主客观质量,但是SAO编码参数存在大量冗余,导致其不能适用于低比特率环境。因此,提出了一种基于时域相关性的自适应采样值补偿算法。通过率失真代价来选择当前帧是否复用最近已编码帧的SAO编码参数,并为每一帧设置一个标志位来标识是否复用。实验结果表明,与HTM原始SAO滤波器算法相比,该算法在主观质量和复杂度基本不变的情况下,3个视频分量的BD-rate平均分别减少1.06%、1.10%和0.99%。因此,所提出的算法能很好地改善SAO的性能。

Although the subjective and objective quality of video can be improved effectively by adopting sample adaptive offset （SAO） filter, it＇ s not suitable for the real-time application for the large amount of parameters in SAO filter. Therefore, a simple but effective algorithm based on temporal correlation is proposed. In coding process, rate distortion （RD） cost is calculated to determine whether the parameters of former coding frame can be used directly by current coding frame, and then a flag bit is used to mark the result. Experimental results demonstrate that the proposed method can reduce 1.06%, 1.10% and 0. 99% for YUV components re- spectively in BD-rate as compared with HEVC-based test model 6. 1 （ HTM 6. 1 ） anchor, with almost the same subjective quality and computational complexity, which can verify the effectiveness of the proposed approach.