H.265/HEVC Video Coding Over Lossy Networks: Flexible or Fixed Mode in One CTU?

Flexible mode selection in one CTU, as one of the new technologies in H.265/HEVC video coding, can attain a high compression ratio. However, it may decrease the end-to-end PSNR for video communication over lossy networks. In this paper, we analyze the flexible prediction mode for the CTU in H.265/HEVC and find it causes more error propagation than fixed prediction mode in sub-pixel interpolation. Then, fixed mode-based error resilient encoding (FERE) is proposed to decrease the influence of error propagation. Experimental results show that the proposed FERE algorithm improves video quality by up to 5.56 dB compared with the baseline algorithms.