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.