A fast image coding algorithm using variable-rate mean-match correlation vector quantization

A variable-rate mean-match correlation vector quantizer, MMCVQ, is presented as an alternative vector quantizer of side-match vector quantizer (SMVQ) for fast image encoding. Before encoding, the mean values of all codewords are computed, then a sorted codebook is obtained according to the ascending order of the mean values of codewords. During the encoding stage, high correlation of the adjacent image blocks and mean values of the current processing vector and codewords are considered to obtain the state-codebook of the current processing vector. The experimental result shows: (1) the encoding time of MMCVQ is almost 7 times faster than that of full-search VQ and almost 42 times faster than that of SMVQ, and (2) compared with SMVQ, the encoding quality will increase more than 1 dB in PSNR, and (3) the bit rate of MMCVQ is smaller than that of SMVQ by 0.03-0.05 b/pixel