Abstract: Recently, parallelization of geometry correction on remote-sensing image has become an important research area. But there exist some problems in current parallel algorithms: They do not have the capability of load-balance, or the global computation is very large. What's more, local operation is time-consuming. Therefore, a parallel geometry correction algorithm based on dynamic division-point computing, called PIWA-DDC, is proposed. By means of the LogP model, it is deduced that PIWA-DDC has good scalability. The results from experiments on MPP show that this algorithm has good capability of load-balance and high efficiency of geometry aberration processing.