Abstract
The High Efficiency Video Coding (HEVC) adopts a hierarchical quad-tree based coding unit (CU) partitioning structure, which allows to recursively split a block into four equally sized blocks. At each depth level, there are up to 35 intra prediction modes. Therefore, enormous computational complexity is introduced due to the recursive Rate-Distortion Optimization (RDO) process on all possible depth levels and prediction modes to find the optimal CU partitions. In this paper, a fast mode decision algorithm is proposed for HEVC intra coding. A logistic regression classifier is introduced to early terminate CU splitting decision process, which is formulated as a binary classification problem. An offline training scheme is applied to obtain a set of efficient coefficients for the logistic regression classifier. Efficient and computationally-friendly features are extracted based on an F-score approach for different QPs and CU depth levels. The experimental results show the proposed algorithm can reduce computational complexity by 55.5% on average with 1.29% Bjontegaard Delta bitrate (BDBR) increase for various test sequences under “encoder_intra_main” condition compared with that of HEVC reference test model.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call