Title

Three-Dimensional Embedded Color Image Coding Using Zerotrees of Wavelet Coefficients Applied to Visible Human Dataset

Author

Yi Mu

Date of Award

2005

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computing

First Advisor

Beddhu Murali

Advisor Department

Computing

Abstract

A lossless to lossy medical color image compression algorithm based on three-dimensional wavelet transform and zerotree coding (EZW) is presented. The algorithm (3D-CEZW) efficiently encodes color image volume by exploiting the dependencies in all three dimensions, decorrelating the original RGB color space to YCC color space, while enabling lossy and near lossless compression from the same bit stream and maintaining the fully embeddedness required by color image encoder. An application to Visible Human datasets shows that the 3D algorithm produces 3-5dB higher PSNR than 2D algorithm. Compared to Adaptive Arithmetic coding, 3D-CEZW is superior in lossless compression which is desired for many medical applications.