Multi-Resolution Structure Color Image Compression Using DWT and VQ

Abstract

Quad-tree structure is an excellent algorithm for block based image processing, especially image compression. The structure divided an image block into homogenous blocks of size 16×16 , 8× 8 , 4× 4 , and non-homogenous blocks of size 4× 4 . The bottom-up Quad-tree structure is very suitable for very low bit rate image compression of high performance using DWT and VQ. The multi-resolution decomposition by Discrete Wavelet Transform (DWT) is an efficient analysis tool for image processing. In general, 4-level decomposition is suitable for color image compression. In the paper, an adaptive clustering skill for generating multi-resolution codebooks by vector quantizing wavelet transformed sub-images of four level of decomposition is proposed to color image compressed. The quality of the recover is varying reconstructed from a multi-resolution codebook. Experiment results shown that the proposed scheme can achieve very low bit rate with high performance for color image.