A Steady Corner Detection of Gray Level Images Based on Improved Harris Algorithm

This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. Accurately and stably detecting corners is needed for good matching, even for image with rotation and noise. As a well known excellent corner detection method, Harris´ technique has the shortcoming of lacking accuracy and stability for images with rotation and noise. This paper improves Harris´ method for better locating corners as follows: the square neighbor of original method is replaced by round neighbor; the intensity change rate of horizontal and vertical directions in the square neighbor is replaced by the statistical intensity change rate according to different angles and radius. Experiments indicate that our method has better corner localization for synthetic, standard and natural images with rotation and noise.