An Improved Interpolative Vector Quantization Scheme Using Visual Pattern (IIVQ)

In this paper, a new image compression scheme by introducing visual patterns to interpolative vector quantization (IVQ) is proposed. The goal is to introduce visual patterns on designing the codebook, where only removed details that contain visual patterns and their original counterparts as pairs are trained. In the proposed scheme first input images are down-sampled by ideal filter. Then, the down sampled images are compressed lossly by JPEG and transmitted to the decoder. In the decoder side, decoded images are first up-sampled to the original resolution. The codebook is designed using LBG algorithm. One of main contributions in this paper is the Experimental results show: (a) visual pattern blocks are easy to form clusters than original blocks; (b) the proposed scheme achieves much better performance over JPEG in terms of visual quality and PSNR.