Curvature statistic corner detection

Corner detection is a basic task in the field of image processing. This paper proposes a new corner detection method based on the curvature computing, called Curvature Statistic (CS) method. The edge points which have the maxima of absolute curvature are defined as corners, so curvature of every edge point is computed in CS. The first step is to use Canny edge detector to extract edges of original image, then compute the curvature of every edge point using a statistic method in the neighborhood. When we get the curvature of the point, we can judge whether it is a corner point or not. This method is very robust to noise. In the section of the experimental comparison we can see that CS performs good, and that most corners are detected and the localization of each corner point is more accuracy.