作者: S.Y. Hu , M.C. Chen , A.N. Wilson
DOI: 10.1109/ISCAS.1997.622113
关键词: Data compression 、 Linde–Buzo–Gray algorithm 、 Quantization (image processing) 、 Mathematics 、 Motion compensation 、 Computational complexity theory 、 Motion vector 、 Rate–distortion theory 、 Rate–distortion optimization 、 Algorithm
摘要: A fast optimization algorithm that jointly selects motion vectors (MVs) and quantization levels based on a rate-distortion (R-D) criterion is developed for motion-compensated video coding system. In contrast to an optimal R-D algorithm, which requires performing DCT scalar operations, can be computationally excessive practical applications, the proposed achieves significant complexity reduction by efficiently pruning MV candidates modeling residual bit-rate overall distortion with simple polynomial functions of level given vector. Experimental results indicate our maintains compression performance close while keeping computational conventional heuristic approaches.