Improved Rectangle Template Matching Based Feature Point Matching Algorithm

Abstract

Feature point matching is an essential section in image matching. Traditional feature point matching algorithm uses merely the information of feature points such as the descriptors, and on some occasions this may lead to a loss of feature points or even mismatches. This paper proposed an improved feature point matching algorithm based on rectangle template matching. Provided with 2-dimensional coordinates of feature points, a novel parameter named "inner-point index" is calculated and used to extract templates from the input image. On the basis of template matching between the input image and the other image, the process of feature point matching is carried out between templates and the matched areas, using Brute Force matching algorithm. Experimental results show that the proposed algorithm can obtain more correct feature point matches, thus enhances the accuracy of feature point matching.