基于ISA-DWT的MR图像压缩新算法

本研究提出一种新的ISA-DWT（整数到整数的形状自适应离散小波变换）的核磁共振（MR）图像压缩算法,对变换后的系数采用适合于形状自适应离散小波变换的修改的SPIHT算法进行编码,并增加上下文自适应算术编码以提高其压缩性能.本研究算法可将前景区域和背景区域生成的压缩码流完全分离,且小波域的系数个数和图像域的个数相同.实验结果表明,对于MR图像,本研究算法的压缩性能明显优于JPEG2000中的最大位移算法.

A New Coding Scheme for MRI Using Integer-to-integer Shape Adaptive Discrete Wavelet Transform

This paper proposed an integer-to-integer shape adaptive discrete wavelet transform(ISA-DWT) coding scheme for the MRI.The well-known set partitioning in hierarchical trees(SPIHT) was modified to generate the compression bit streams,and the context-adaptive arithmetic coding was added to enhance the compression performance.The compression bit streams of the foreground and the background image were completely separated and the number of coefficients after ISA-DWT was identical to the number of the pixels of the image.Experimental results indicated that the proposed approach significantly outperformed the maxshift method of the JPEG2000 for MRI.