| Summary: | 碩士 === 國立成功大學 === 電機工程學系 === 103 === High Efficiency Video Coding (HEVC) is the newest video coding standard. It provides the better compression performance compared with the existing standards. HEVC adopts the quad-tree structure which allows recursive splitting into four equally sized nodes, starting from the Coding Tree Unit (CTU). The quad-tree structure causes the better compression efficiency, but it requires the higher computational complexity.
In order to reduce the computational complexity, we propose a fast CU size decision algorithm. The proposed algorithm consists of adaptive depth range and early pruning test. First, we use adaptive depth range instead of fixed depth range for a CTU encoding. Then, for each CU, the early pruning test is performed at each depth level according to Bayes rule based on the full RD costs.
Compared with the HEVC test model 12.0 (HM 12.0), experimental results show the proposed method achieves the reduction of encoding time by 60.11%, the increment of bitrate by 2.4%, and 0.1 dB Y-PSNR loss, on average.
|
|---|
