An optimized vector quantization for color image compression

Date

1998-05

Authors

Kompella, Sastry V S

Journal Title

Journal ISSN

Volume Title

Publisher

Texas Tech University

Abstract

Image Data compression using vector quantization (VQ) has received a lot of attention in the recent years because of its optimality in rate distortion and adaptability. A fundamental goal of data compression is to reduce the bit rate for transmission or data storage while maintaining an acceptable fidelity or image quality. The combination of subband coding and vector quantization can provide a powerful method for compressing color images. Most of the existing VQ algorithms however suffer from a number of serious problems like long search process, codebook initialization, getting trapped in local minima, etc. This work investigates the development of an image compression algorithm using a variable block size vector quantization technique for generation of optimal codebook by employing a neuro-fuzzy clustering approach to ensure minimum distortion. Each color image is decomposed into R, G, and B color planes prior to application of wavelet transform and vector quantization to each color plane. Each color plane is preprocessed by performing multiresolution wavelet decomposition. The multiresolution nature of the discrete wavelet transform is utilized to decompose the images into more directionally decorrelated sub-images, which are more suitable for quantization and coding. Vector quantization is performed on each of the subimages at different resolutions and a multiresolution codebook scheme is utilized. This new approach to image compression facilitates generation of an improved globally optimal codebook, and a simpler search scheme. Finally, the codebooks generated from the three encoded color planes are entropy coded for obtaining higher compression at minimum distortion. Each color plane codebook is decoded and the reconstructed color planes are combined to form the final reconstructed image. The reconstructed images are compared with those of other standard compression algorithms, in terms of Mean square error (MSE), and Peak signal-to-noise ratio (PSNR).

Description

Keywords

Signal processing -- Digital techniques, Data compression (Telecommunication), Image processing -- Digital techniques, Coding theory, Image processing -- Image quality

Citation