基于ISA-DWT的MR图像压缩新算法

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

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。

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.