Fast CU Size Decisions for HEVC Intra Frame Coding Based on Support Vector Machines

High Efficiency Video Coding (HEVC) achieves themost significant coding efficiency compared with all the existingvideo coding standards. However, the Intra encoding complexityis increased dramatically since the complex recursive searchalgorithm for the coding unit (CU) size decisions. In this paper, afast CU size decision algorithm based on Support Vector Machines(SVM) is proposed to further alleviate the encoder computationload. Firstly, some effective image features are extracted to judgethe CU complexity. Then, a three-classifier structure of CU sizedecisions is established based on SVM. The experimental resultsshow that the proposed algorithm can reduce 53% intra codingtime on average with 1.2% BDBR loss on average.