Fast HEVC intra coding algorithm based on Otsu's method and gradient

High Efficiency Video Coding (HEVC), the newest video coding standard established, aims to significantly improve the compression performance compared with H.264/AVC and other existing video coding standards. In the intra coding of HEVC, this improvement is achieved by adopting the coding tree unit (CTU) structure and providing up to 35 prediction modes for each prediction unit (PU). However, the computational complexity is greatly increased due to these new techniques. In this paper, we propose a fast coding unit (CU) size decision and PU mode decision algorithm for HEVC intra coding to reduce the time consumption. Since CU sizes are highly correlated to the texture detail and variance information of the largest coding unit (LCU), we introduce the Otsu's method to measure the texture complexity of each LCU and skip some CU depth levels. Meanwhile, as the prediction modes always correspond to gradient directions, we apply improved Sobel masks to measure gradient directions of PUs and reduce some candidate intra modes. Experimental results show that our proposed algorithm can save 33% time consumption on average with negligible loss of coding efficiency.