基于MPI和OpenCV遥感图像匹配的并行实现  

Parallel implementation of remote sensing image matching based on MPI and OpenCV

作　　者：崔丹丹[1,2] 张耀南[1,3] 赵国辉[1,3] 

机构地区：[1]中国科学院寒区旱区环境与工程研究所,兰州730000 [2]中国科学院大学,北京100049 [3]甘肃省高性能网格计算中心,兰州730000

出　　处：《计算机应用》2014年第A01期203-205,共3页journal of Computer Applications

基　　金：国家基础科学人才培养基金冰川冻土特殊学科点资助项目(J1210003;J0109);国家自然科学重点基金资助项目(91125005/D011004)

摘　　要：灰度匹配是遥感图像处理中一项重要技术,匹配方法虽然精度高,但计算量大,时间长。针对这一问题,介绍了遥感图像灰度匹配的并行策略,基于消息传递接口(MPI)和开源计算机视觉库(OpenCV)实现了归一化积相关灰度匹配并行算法,并依此算法来验证遥感灰度匹配的并行性能。实验结果表明,该算法能获得良好的加速比且并行效率大于1,说明基于该算法的遥感图像匹配可以显著提高遥感图像处理的效率。Gray matching is an important technology in remote sensing image processing. Although this method has high precision, it requires a long computation time because of computation-intensiveness. To solve this problem, this paper introduced the parallel strategy of image gray matching. It implemented the normalized product correlation algorithm based on MPI and OpenCV to validate the performance of parallism . The experimental results show that it obtains good speedup and parallel efficiency of greater than l, indicating that the proposed method can significantly improve the efficiency of remote sensing image processing.

关 键 词：遥感图像处理 图像匹配 消息传递接口 开源计算机视觉库 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]