An efficient HEVC multi-rate encoding system based on ×265
Chunlei Cai, Shibo Yin, Xiaoyun Zhang, Zhiyong Gao · 2016
Multi-rate encoding using HEVC is in great demand but with huge complexity. Although ×265 is an efficient HEVC encoder, it has not been optimized for multi-rate encoding and its parallelism is still not enough on multi-core processors. In this paper, we deeply investigate the coding correlation including depth and prediction mode between HEVC streams at multiple rates. Then an efficient HEVC multi-rate encoding system is proposed which consists of depth sharing and prediction modes pruning based on Naive Bayes Classification. Experimental results show that the proposed multi-rate encoding system can reduce 44.68% of encoding time with 0.016dB BD-PSNR performance improvement and much higher CPU utilization compared with independently ×265 encoding multiple times.