红外与激光工程, 2004, 33 (5): 496, 网络出版: 2006-05-25
MPI并行计算在图像处理方面的应用
Applications of MPI parallel-computing on image processing
消息传递接口 并行计算 灰度相关 集群 快速傅里叶变换 Message passing interface Parallel-computing Intensity-based correlation Cluster FFT
摘要
介绍了MPI并行计算的基本概念和并行计算集群的实现,着重论述了图像处理算法的并行实现方法.列举了提高并行算法效率的一些措施.在并行计算集群上实现了灰度相关匹配算法和快速傅里叶变换算法,对它们的实验结果进行了分析和对比,通过实验数据说明了并行计算通信量和并行计算效率的关系,提出了并行计算在图像处理方面的适用范围.
Abstract
The concept of parallel-computing based MPI and the implementation of parallel-computing cluster are introduced. The parallel way of image processing arithmetic is presented. Some methods of impoving parallel efficiency are put forward. A matching arithmetic based on intensity-based correlation and the FFT arithmetic are implemented and run on the parallel-computing cluster. The results are analyzed and compared. The expermental data show the relationship of computing efficiency and the quantities of communication. The feasibility scope of parallel computing in image processing is put forward.
吕捷, 张天序, 张必银. MPI并行计算在图像处理方面的应用[J]. 红外与激光工程, 2004, 33(5): 496. 吕捷, 张天序, 张必银. Applications of MPI parallel-computing on image processing[J]. Infrared and Laser Engineering, 2004, 33(5): 496.