嵌入式零树小波压缩和解压缩的并行化算法  

Parallel algorithm for image copression and decompression based on EZW

作　　者：韩丽洁[1] 李文[1] 田晏嘉[1] 

机构地区：[1]内蒙古大学计算机学院,呼和浩特010021

出　　处：《计算机应用》2009年第B06期209-212,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60303033);内蒙古大学513人才计划项目;内蒙古大学自然科学基金资助项目

摘　　要：嵌入式零树小波压缩算法是图像压缩技术中有效的压缩算法,但其压缩时间较长。对该算法进行了研究,并在多核机群系统下实现了该算法的并行算法,提高了算法的性能。实现了MPI和MPI+OpenMP两种并行算法,并将串行算法、MPI并行算法与MPI+OpenMP并行算法进行比较。结果显示,随着数据量的增多,MPI并行算法和MPI+OpenMP并行算法相对于串行算法的运行效率都有明显提高,其中MPI+OpenMP并行算法的效率更好。Embedded Zerotree Wavelet （EZW） is an effective algorithm for image compression. However, image compression is a time-consumlng process. The parallel algorithm of EZW was implemented for improving performance on multicore cluster. The parallel algorithm using MPI and MPI ＋ OpenMP was developed. Comparision of serial algorithm and parallel algorithms shows that the efficiency of two parallel algorithms has been enhanced obviously with input data increasing. Furthermore, the algorithm using MPI ＋ OpenMP is more effective than that using MPI.

关 键 词：MPI MPI+OPENMP 嵌入式零树小波编码 

分 类 号：TP338.6[自动化与计算机技术—计算机系统结构]