In this paper, a feature matching strategy is developed. It is realized by introducing a function whose independent variable is the match matrix, which describes the correspondence of the features. It combines spatial relations and feature similarity organically and makes sure that its global maximum can be reached when the sensed image is aligned with the reference image completely. Thus the feature correspondence can be estimated by finding the maximum of the function. The branch-and-bound strategy is employed in the integer programming problem. A lot of real images are used to demonstrate its performance. Compared with some existing methods, it is automated, robust, and has the highest accuracy.