A robust corner detector based on curvature scale space and harris

The traditional corner detector based on Curvature Scale Space is sensitive to noise, and has its limitations in detecting certain X corners. On the basis of studying CSS corner detector, this paper proposes a refined corner detector, where we use both CSS and Harris corner detector to detect corners, and then use the clustering method to analyze the candidate corners and assign the clustering centers as the coordinates of detected corners. Experiment results show that the proposed method offers a robust and effective solution to eliminate falsely detected corners and redundant corners.