基于多小波域新型空间方向树的SPIHT改进算法

An improved SPIHT algorithm based on a new type spatial orientation tree in multiwavelet domain

为了更好地利用图像多小波域系数的相似性进行编码,提出了一种基于多小波域新型空间方向树的SPIHT改进算法。在SPIHT改进算法中首先定义了一种新型空间方向树把多小波域中不同子带的相似系数和每个子带中各个子块的相似系数组织在一起,然后根据树的结构改进了算法中分裂排序过程。实验结果表明,SPIHT改进算法与原来的SPIHT算法相比,在相同的压缩比下可以获得更高的峰值信噪比。

In order to fully exploit the self- similarity of image coefficients in multiwavelet domain, an improved SPIHT algorithm based on a new type spatial orientation tree in multiwavelet domain is presented. In this algorithm, a new type spatial orientation tree in multiwavelet domain is defined. This tree not only connects the self- similarity coefficients between different subbands together, but also connects the self- similarity coefficients between different blocks of each subband together. Experiments show that the improved SPIHT algorithm could provide higher PSNR value than the original SPIHT algorithm at the same compression rate.