A Modified Embedded Zerotree Wavelet Algorithm for Medical Image Compression

The Embedded Zerotree Wavelet (EZW) algorithm is based on successive approximation of wavelet coefficients at different subbands. A new analytical and numerical based investigation on the dependence of this process on the probability density function of the data to be approximated is presented. We have shown that the value of threshold-weighting factor in S.A process plays an important role in the performance of the EZW algorithm. It has been shown that the shape parameter in the Generalised Gaussian function, a model for the pdf of the wavelet coefficients, is related to the performance of the EZW algorithm. The results of the above investigations have been used to make a great improvement in the performance of the EZW algorithm for compression of medical images.