适于硬件实现的3D SPIHT算法优化设计  

Optimal Design of the 3D SPIHT Algorithm Suitable for Hardware Implementation

作　　者：魏本杰[1] 刘明业[1] 张晓昆[2] 周艺华[1] 

机构地区：[1]北京理工大学计算机科学技术学院,北京100081 [2]北京电子科技学院计算机系,北京100070

出　　处：《北京理工大学学报》2006年第2期125-129,共5页Transactions of Beijing Institute of Technology

基　　金：国家部委预研项目(41308010408)

摘　　要：通过对一种可精确控制压缩率的三维多分等级树(3D SPIHT)编码算法的分析,为提高编码速度,改进了3D SPIHT的算法结构,简化了动态分配内存的链表结构,采用位图映射的方式固定分配内存,可减少内存86.6%,无需进行硬件难以实现的链表操作.变三维地址为线性地址,并在预处理时建立子代最大值映射表,减少了零树判断的时间.实验结果证明,与原算法相比,当压缩比小于26.7时,优化后算法的速度提高了10倍以上,为进一步的VLSI设计奠定了基础.By analyzing three dimensional set partationing in hierarchical trees（3D SPIHT） coding algorithm, which can accurately control the compression ratio, this paper improves the architecture of 3D SPIHT to speed up video coding, then simplifies the structure of linking lists allocating memory dynamicallY, and takes the of bitmap way to allocate memory fixedly without doing the difficult operations for the hardware design. This can reduce the memory by 86.6 %. Translating 3D address into linear address has been done and a mapping table is built for searching the maximum value of the descendants, which can reduce the time of zerotree judgement. Compared with the original algorithm, the experimental results prove that the optimal algorithm can increase the speed more than ten times when the compression ratio is smaller than 26.7, that builds the basis for further VLSI design.

关 键 词：3D离散小波变换 3D多分等级树 位图映射方式 线性地址 子代最大值 

分 类 号：TN911.73[电子电信—通信与信息系统]