Edge and mean based classified side-match finite-state vector quantization for image compression

摘要

A new side-match finite-state vector quantization (SMVQ) combined with an efficient classifier called edge and mean based classified side=match finite-state vector quantization (EM-CSMVQ) is proposed in this paper. EMCSMVQ exploits the advantages of classified vector quantization (CVQ) and SMVQ,where the edge characteristics and the mean properties of the image blocks are used by the classifier. In this work, the detection of edge can be easily performed by a simple method called intuitive edge extraction (1EE). Simulation results show that the improvement over SMVQ and VQ can be up to 2.91 dB and 4.32 dB, respectively, at lower bit rate. Moreover, when EMCSMVQ is compared with VSMVQ, it takes fewer bits and obtains beter image quality.