Scene Matching Algorithm Based on Corner Feature Detection

摘要

How to match two images obtained from different sensors or under different imaging conditions is a difficult problem for traditional matching methods. To obtain the higher matching probability, improving the similarity between the reference image and the real image is one of the most effective means. In this paper, firstly, we mainly consider the corner feature extraction to maintain the steadiness feature of the image. At the same time, we detect the edge of the images. These two kinds of features were combined to the image matching process. And then, we present a matching technique based on fast Distance Transformation. This paper also introduces a 'coarse to fine' matching strategy. Firstly, to find the probable matching positions, we match the two corner images with the matching method of the distance transformation. Lastly, to find the exact matching position, only at these probable positions we match the two corresponding edge-enhance images with the co-correlation of the edge intensity. The experiments show that this method is effective.