基于统计模型的三线阵CCD影像并行匹配算法

Parallel algorithm of CCD image matching based on statistical shape models

在基于统计方法实现CCD影像匹配算法的基础上,提出了CCD影像匹配计算任务的分解方法和计算任务的调度策略,进一步构建了基于统计模型的三线阵CCD影像匹配算法的并行计算方案。并通过实验进行了验证,结果表明,对于CCD影像该并行匹配算法获得了良好的加速比,可扩展性好。适用于超大规模的三线阵CCD影像中同名像点的匹配计算。

Fast CCD image matching is a key technology to achieve lunar three-dimensional reconstruction in China′s lunar exploration project.In this paper,according to some available researches,a parallel CCD image matching algorithm based on statistical model has been presented and implemented.In the algorithm,we study the load balancing problem for CCD image matching at first,and then we establish tasks decomposition method,which uses image data division.At last,we build a task scheduling strategy for CCD image matching algorithm,which not noly reduces the communication consumption,but also achieves better load balancing.Experiments show that the parallel matching algorithm for CCD image has good speedup and scalability.The algorithm is suitable for the ultra-large-scale CCD image matching.