| The technique of image compression is the important area of multimedia research. The EZW (embeded zerotree wavelet) is deemed to be the most efficienct algorithm for image compression. Image compression is a time-consuming process, therefore a more effective method should be found for extending the application of EZW. In this thesis, multi-core is combined with the EZW. While good compression effect is obtained, the quicker compression time is getted. First of all, how to actualize serial EZW algorithm is introduced. Secondly, how to actualize parallel algorithm on multi-core cluster is intruduced. Two methods for parallel algorithm are given in this period. The first method involves application of MPI programming mode, which realizes parallelization among cluster nodes. The second employs MPI +OpenMP hybrid parallel programming mode, in which MPI still actualize the parallelization among cluster nodes, but OpenMP actualize the parallelization in node. At last, the effect of serial program is analyzed and the performance of serial algorithm is compared to parallel algorithm. The result shows that the efficiency of image comeback is very well. And as the number of dealing with data increaseing, the efficiency of two parallel algorithms is improved obviously. Furthermore the effect of MPI+OpenMP program is better than that of the MPI program. |
PDF Full Text Request