Multistage vector quantization based on simulated annealing for image coding

Multistage vector quantization (MSVQ) can achieve very low encoding and storage complexity in comparison to unstructured vector quantization. However, the conventional stage-by-stage design of the codebooks in MSVQ is suboptimal with respect to the overall performance measure. The generation of codebooks from a set of training vectors using simulated annealing (SA) can converge to the globally optimal codebook in finite time. In this paper, SA is combined with MSVQ for optimizing the overall performance. The algorithm of multistage vector quantization based on simulated annealing (MSVQ-SA) for image coding is presented. Experimental results show that MSVQ-SA technique results in significantly better quantizer performance than that obtained with SA and MSVQ.