Application of VQ to Optimal Transform Coding

Abstract

In this paper, an optimal adaptive transform coding system based on VQ is proposed. In the proposed approach, the K-means clustering is used to find representative image blocks through which VQ-based masks are found. The set of VQ-based masks is then applied in the selection of transform coefficients. It is shown that the set of VQ-based masks is optimal in the sense of minimum average energy loss. Next, the set of optimal VQ-based masks is applied in JPEG. The modified JPEG with optimal VQ-based masks is termed as OJPEG. Simulation results are provided to justify the optimality in VQ-based masks. Besides, the proposed OJPEG is compared with JPEG in terms of PSNR and bit rate. It indicates that the proposed OJPEG is able to trade little PSNR with significant bit rate reduction when compared with JPEG.