Efficient variable rate vector quantization using quadtree segmentation

Abstract

Variable rate vector quantization (VQ) using greedy tree growing algorithm has been used in image compression. In previous work, images have been partitioned into fined size blocks and therefore correlation existing in large variable size blocks cannot be exploited. In this paper, a new and efficient variable rate vector quantization scheme using variable block sizes is introduced. Simulation results show that the proposed scheme which is based on variable block size quadtree segmentation can achieve better performance than the fixed block size variable rate VQ.