Research About Fast Rotation Invariant Template Matching Method

With the development of modern computer science and technology, the use of computer image processing operations have become more common, and played a more and more important role in several parts in our daily life. Image matching issue is the basic issue in image processing field. And it is widely used in Computer Vision, Aircraft Navigation and Target Recognition and Tracking.Image matching algorithm is a process which searches the corresponding pattern by a known pattern. So far none of the current matching algorithms are universally applicable for all cases. The existing image matching algorithms have shortcomings either in speed or in accuracy or robustness. Therefore, a new template matching method which solved the questions above is required.In this paper, the author did a lot research about those questions. At first, the author compared the ways of work of traditional matching methods in order to find out the reason why they can not match when images rotated. Secondly, the author compared some fast searching methods and rotation invariant methods and found the advantages and disadvantages of these methods. Finally, the author proposed a new rotation invariant image matching method.In this theory, the image matching process is separated into two steps. Firstly, it does a coarse matching for finding the candidate matching points fast. Secondly, it does a precise matching which uses Pseudo Zernike moment to get accuracy result.The author offered the implementation of the new matching method in the paper. After that the author did a lot experiment to verify the accuracy and validity, and gave a detailed explanation of the speed, rotation independence and robustness. The experimental result says that the proposed algorithm has the ability of matching image rotation invariantly in a high speed, in the mean while keeps robust of noisy in the reference images.