基于二进索引树的算术编码在SPIHT算法中的实现  

Implementation of Arithmetic Coding Algorithm Based on Binary Index Tree in SPIHT

作　　者：汪国有[1] 翁传强[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,武汉430074

出　　处：《计算机与数字工程》2005年第1期1-4,共4页Computer & Digital Engineering

基　　金：国家自然科学基金资助 (6 0 372 0 6 6 )

摘　　要：多级树集合分裂算法 (SPIHT) [2 ] 是在嵌入式零树小波编码算法 (EZW) [1] 的基础上改进的性能更优的算法。其子集合分裂策略是如此的有效以及显著信息是如此的紧凑 ,以至未编码的二进制位流也能获得同EZW一样的甚至更好的效果。虽然引入算术编码可以略微提高峰值信噪比 (PSNR) ,但同时也增加了算法的计算复杂度。在实现SPIHT算法时 ,通过基于 2× 2基元的方式来组织LIS和LIP表 ,大大地提高了SPIHT算法的运算速度 ;但同时算法的重建图像质量也有大幅度的下降。基于上述事实 ,本文引入了基于二进索引树的算术编码算法。实验表明 ,在SPIHT算法的基础上引入基于二进索引的算术编码 ,可以使得SPIHT算法运算速度大大增加的同时其重建图像质量基本不下降。SPIHT(Set Partitioning in Hierarchical Trees) [2] is an excellent algorithm modified on the basis of EZW(Embedded Zerotree Wavelets) [1] coding algorithm. Its sub-set partitioning policy is so efficient and its significant information is so tight that its output un-encoded binary stream has same or much better result than EZW. Although, introducing arithmetic coding algorithm can get a few PSNR gains, its algorithm complexity has increased at the same time. When implementing SPIHT, organization method based on 2×2 block of LIS and LIP greatly increases its encoding speed. But, the reconstructed image's quality has declined too. Based on the former fact, we introduce the arithmetic coding algorithm based on binary index tree. Experiment shows that by introducing arithmetic coding algorithm based on binary index tree, coding speed of SPIHT has increased greatly and it's reconstruct image quality has little decline.

关 键 词：算术编码 SPHIT二进索引树小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]