High Performance Image Compression Based on Optimized EZW Using Hidden Markov Chain and Gaussian Mixture Model

This paper has proposed a novel high performance image compression method based on Embedded Zerotree Wavelet (EZW), which is a popular image compression coding scheme. Shortening the sweeping path according to the statistical properties of the image is the main issue that this paper has concentrated on. A developed statistical model based on hidden Markov model (HMM) and Gaussian mixture model (GMM) trains and produces the filter-bank coefficients of discrete wavelet transform (DWT). Custom design filter-banks for specific category of images, allow the encoder to pass a shorter route and stop at a closer point. This research proposes a novel image compression method based on an optimized EZW transform for three category of images, from statistically high to low correlated images. In each of these categories, the filter-banks are trained and extracted by their related GMM-HMM statistical models. The final results prove the efficiency of the method, and they are compared against the state of the art methods.