Corner detection algorithm based on Hough transform

郭斯羽,孔亚广,张煦芳

DOI: https://doi.org/10.3321/j.issn:0254-3087.2008.11.035

2008-01-01

Abstract:Corner detection is a basic and important issue in the area of image processing and machine vision. A corner detection algorithm based on Hough transform was proposed. The neighboring window of each pixel in an edge image is investigated iteratively, and Hough transform is applied to find the straight lines from the center of the window. The number of these lines and the angles formed by each pair of adjacent lines are inspected to determine whether the center pixel of the current window is a corner point. For a window with given size, template computable a priori and circular buffer are utilized to boost the efficiency of the Hough transform and the image pixel traversal process. Comparison experiments on a real image base show that the proposed method gives high corner detection rate. The proposed algorithm also outperforms other methods in low false positive rate of the detection and the corner position errors. Compared with the algorithms giving the arm directions of the detected corners, the novel method also shows a higher arm detection rate and accuracy. For applications that are not time-critical, the proposed algorithm gives better overall detection performance than other algorithms do.