Efficient pyramid VQ-based image compression algorithm

In this paper a high-quality and low-computational-complexity compression scheme for still images is presented. The image to be compressed is partitioned into an abstract and a texture image, each of which could be separately processed. The abstract image contains the four most significant bits and the texture on the remaining least significant bits of the original image. In the experiments performed, the proposed compression algorithm has been tested on standard images. The separation of the image into two images enables their parallel compression, improves the PSNR, and reduces the blocking effect. The advantages of this algorithm are a) each block is separately vector quantized without a need for the DCT computation of other neighbouring blocks, B) the processing of data with a half of length reduces the complexity and makes the algorithm faster, and c) the algorithm could be implemented in parallel.