摘要
1 引言小波(Wavelet)变换是20世纪80年代中期出现的新时频域信号分析工具。自1989年S.Mallat首次将小波变换引入图像处理以来,小波变换以其优异的时频局部能力及良好的去相关能力在图像压缩编码领域得到了广泛应用,并取得了良好的效果。其中,由A.Said和W.A.Pearlman提出的SPIHT编码算法被认为是目前国际上最先进的方法之一,该算法以零树集合及其分割排序为基础,不仅结构简单、无需任何训练、支持多码率,而且具有较高的信噪比和较好的图像复原质量,总体性能优于J.M.Shapiro提出的EZW算法。然而,理论分析和实验结果表明,该算法也存在因计算复杂而导致编码速度慢、内存需求量大而导致难于硬件实现等弱点。
In this paper . a new reduced memory zerotree wavelet image coding algorithm is presented. It can be viewed as an improved version of SPIHT algorithm. Unlike SPIHT, our new algorithm uses no lists during encoding/ decoding and revises the SPIHT's sorting strategy. Without the lists, the memory requirement for the VLSI implementation is reduced significantly. And through revising the sorting strategy, the computational complexity is reduced efficiently. The experiment results show that the PSNR values from our coding algorithm are slightly lower than SPIHT's, and encoding/decoding time from ours is greatly economical than SPIHT's.
出处
《计算机科学》
CSCD
北大核心
2002年第8期104-105,共2页
Computer Science
