一种用于图象配准的或图最佳搜索算法

An Optimal Search Algorithm of OR-Graph for Image Registration

本文在分析研究图象配准中的SSDA法的基础上,提出一种用于图象配准的或图最佳搜索算法,该法将窗口图及基准图的子图划分为n块,形成完整的有序或图,使图象配准问题变为用或图最佳优先搜索策略求最短路径问题,实现图象配准。实验结果表明,该算法控制简单,易于实现,比传统的直接算法计算量少,试验效果好。

The difference between two images can be obtained by subtracting one from the other. A practical problem existing in the difference image is the registration noise caused by incomplete cancellation of the unchanged objects in the background if two images are not in perfect spatial registration before subtraction.In order to obtain a correct difference image between two given images, it is necessary to register them first. An optimal search algorithm of OR-Graph for image registration is proposed.For this algorithm, a threshold sequence necessary for the SSDA method is not needed here. In this algorithm, the window image is divided into-n blocks of subimages and a heuristic function is defined so that the problem of image registration becomes one of optimal search algorithm of OR-Graph. Test results show that the algorithm is simple for control and can' be operated with higher speed than the SSDA algorithm in certain cases.