一种基于像素域的改进的SPIHT算法  

An Improved SPIHT Algorithm Based on Pixel Domain

作　　者：聂玉明[1] 邹雪妹[1] 

机构地区：[1]上海大学通信与信息工程学院,上海200072

出　　处：《计算机仿真》2009年第1期209-211,232,共4页Computer Simulation

摘　　要：分层树的集分割算法是一种简单、有效的嵌入式零树编码算法。在算法的基础上提出了一种基于空域的改进方法。首先将图像的偶数裂像素减去相邻两奇数列的均值,形成分裂图像。然后将分裂图像的奇数列提取出来作为主图像,将偶数列提取出来作为幅图像。最后将主幅图像以合适的权值分别进行压缩。解码时,将主幅图像分别解码,再恢复分裂图像,最后再恢复原图像。对标准图像进行测试时,并把结果与原算法进行比较。实验表明,在相同的比特率条件下,重构图像的峰值信噪比远高于原算法。SPIHT algorithm is a simple embedded zerotree coding algorithm with high performance. A method based on pixel domain is presented. First of all, subtracting the mean of neighboring two odd columns from the pixel value of even columns to form the split image. Then, taking the odd columns out of the split image to form main image, and the even columns to form assist image. At last, the main image and assist image are compressed respectively with a proper weight. When needing to show the original image, the main image and assist image are decompressed respectively to form the split image, and reveal the original image at last. This coding method is tested by standard image and the experiment results are compared with original SPIHT. The results show that the PSNR value is higher than that of original SPIHT at the same bit rate.

关 键 词：图像压缩 小波变换 零树编码 分层树的集分割算法 

分 类 号：TN919.81[电子电信—通信与信息系统]