Feature-based Image Matching for Aerial Images

Image edges are extracted with a simple algorithm,and the key points of edges are registered in grids.Then for every key edge point,several matching candidates are searched by finding the peak of correlation coefficients,and the initial matching probabilities are calculated with these coefficients.Last,a relaxation optimization algorithm is used to adjust the matching probabilities.From experiments,it can be concluded that the feature-based image matching can improve the accuracy and robustness.