Texture Characteristic based Fast Algorithm for CU Size Decision in HEVC Intra Coding

The High-Efficiency Video Coding(HEVC) standard adopts hierarchical partitioning technique which classified as coding unit(CU), prediction unit(PU), transform unit(TU) based on quad-tree structure. Also, in order to improve intra predict accuracy, HEVC employed 35 types of extended intra prediction modes. Because of those advanced techniques, HEVC achieves almost double times of compression efficiency compared to H.264/AVC. On the other hand, however, high computational complexity derived as a critical problem. In this paper, we present a fast CU size decision algorithm based on texture information to reduce the complexity of Intra mode in HEVC. Experimental results show that the proposed algorithm accomplishes substantial encoding time saving up to 69.5% and averagely 50% with a negligible drop in BD-PSNR and BD-BR. It only shows 0.025dB loss of Peak Signal-to-Noise Ratio (PSNR) and 0.87% increase on bit-rate compared with original HEVC model (HM).