The Reduction of VQ Index Table Size by Matching Side Pixels

Abstract

The vector quantization (VQ) technology is applied to compress an image based on a local optimal codebook, and as a result an index table will be generated. In this paper, we propose a novel matching side pixels method to reduce the index table for enhancing VQ compression rate. We utilize the high correlation between neighboring indices, the upper and the left of the current index, to find the side pixels, and then reformulate the index. Under the help of these side pixels, we can predict the adjacent elements of the current index and then partition the codewords into several groups for using fewer bits to represent the original index. Experimental results reveal that our proposed scheme can further reduce the VQ index table size. Compared with the classic and state-of-the-art methods, the results reveal that the proposed scheme can also achieve better performance.