A complexity-based adaptive tile partitioning algorithm for HEVC decoder parallelization

This work presents an efficient adaptive tile partitioning algorithm for HEVC decoder parallelization, which divides a video frame into tiles that are processed independently and concurrently on multi-cores. The proposed algorithm is designed to not only increase decoding performance and energy efficiency, but also enhance the video quality by efficiently balancing the workloads of tiles among cores. Experimental results show ~15% improvements of decoding speedup and ~12% energy savings on a 16-core system. Furthermore, it provides better PSNR than other algorithms during experiment.