Image Matching Method And Application

Image matching is a very important research of image processing.Because of the variance of taken time,camera angle,natural environment,a variety of sensors and sensors' own reasons, images are not only affected by noise impact,but also contain serious grayscale distortion and geometry aberration。How to achieve a algorithm with high-prcision,high match percentage,strong robutness become a pursuit of the people.This artile studies on the principles and related theory of image matching,especially analzes the current hot and challenging algorithm in the areas of characteristics point matching—SIFT.In order to reduce mismatches in SIFT,this paper presents two methods:In one way,we propose a method for image matching by using Independent Component Analysis to lower the dimension of feature vectors;In another way,the image is rotated by the point's dirction and uses templates matching after SIFT's processing. Images of geometrical distortion are tested and the results show that the two proposed method will raise the ratio of correct.In the end the principle and composition of the tobacco sorter are introduced, together with the analysis of the principle of this system's identifying foreign materials for using image matching.The result of test shows that it can identify and eliminate foreign materials mixed in tobacco effectively,thus,the quality of tobacco is ensured.