Study and Realization of Large Rotation Angle Matching Images Algorithm Based on Charateristic Matching

Combining the advantages of SIFT charateristic matching and pyramid image matching, this paper aims to solve the matching problem of automatic rotation with large rotation angle in aerial photographing. To start with, match the images at the highest pyramid level with SIFT characteristic matching method. Then use the RANSAC algorithm to combine the results from SIFT method. Before matching, predict the initial location of the matching points by using the estimated rotation angle and then make rotation compensation of the matching window images. Last, validate the method by carrying out an aerial digital imagery with large rotation of a certain area.