基于SSIM和SAD的AVS帧内预测模式改进算法

摘要

In the mode selection process of AVS intra prediction, Sum of absolute differences ( SAD ) is used as distortion measurement and choose the current block prediction mode from the i_sad cost, but this measure is proved not to be fit in with human visual ( HVS ) well. Referring to AVS, the structural similarity ( SSIM ) is used in this paper, Its algorithm is relatively simple, and it has good accuracy, so it is better than traditional methods such as PSNR . The SSIM and SAD are put foward as an intra prediction mode selection standard . Results show that the proposed algorithm can achieve better image quality and the better coding efficiency.%在AVS帧内预测模式的选择过程中,以绝对误差和(SAD)作为失真度量,通过相应的i_sad代价函数进行选择,但是这种方法被证明不能很好地符合人眼视觉(HVS).由于结构相似度(SSIM)算法相对简单,准确性好,优于传统的PSNR等方法,因此提出了将SSIM和SAD作为帧内预测模式选择的标准.实验表明,将SSIM和SAD用于AVS帧内预测模式选择,获得了比原AVS更好的重建图像质量和更高的编码压缩率.