Parallel transform domain Wyner-Ziv video encoding scheme for standard definition video

摘要

In this paper, we propose a new structure for the transform domain Wyner Ziv coding to achieve fast and light-weight encoding of standard-definition (SD) video. The proposed scheme divides each picture into multiple tiles which do not have computational dependencies between each other in order to facilitate parallelization. The proposed parallelized encoder is implemented using OpenMP API. In addition, a bit-plane packing parallel LDPCA encoding scheme is also applied for each tile. This scheme does not depend on specific hardware architecture. Compared to MPEG-4 AVC/H.264 INTRA (GOP=2, 8) and ×264 encoder without hardware acceleration, the proposed scheme shows faster encoding by 1.5 ∼ 1.9 and 2.7 ∼ 4.7 times, respectively.