Study Of Image Matching Algorithm

The purpose of Image matching is to find similar couples between two or more images by using a certain matching algorithm, the matching images are from the same object, often based on different time, different perspectives or different size. now this technology is very important and widely used in many fields, such as Computer Vision, Environmental-monitoring, Object Recognition and Tracking, Optical and Radar Tracking. The complexity of the image brings difficulties to the matching process, So selecting a suitable matching algorithm is the key to real-time matching. Image matching methods cover two categories: one is based on gray, the other is feature. The former is based on gray value of matching images, but the calculation is very large; The latter is not sensitive to noise, and can be used in the real-time match.Firstly, two typical image matching algorithms are described; Then two new improved algorithms are given based on the proposed algorithms; Finally, we use programming tools to implement the proposed algorithms, and analyze the matching results to discuss these algorithms.The main contents of the dissertation are as follows:(1) Introduction to classical matching algorithms. Including two algorithms: Template matching and SIFT matching, introduced the basic ideas of these two algorithms, their own implementation and process.(2) Introduction to improved matching algorithms. In this paper, I described the basic ideas of improved algorithms, the process of their implementation, and compared new improved algorithms with previous ones.(3) System Construction and Implementation. At the end of my paper, I introduced the programming tools (including VC++, MATLAB and its own graphical user interface GUI), software interface, key code, and test data analysis.