Corner Detector Algorithm for Colour Image with Highlights and Shadows

Traditional curvature scale space (CSS) corner detection algorithms did not differentiate various physical causes of image differential structure, therefore these algorithms could not detect corners correctly in color image with shadows and highlights. According to the problem above mentioned, dichromatic reflection model was applied to differentiate image differential structure and multi scale localization and multi scale detection algorithm was improved. Finally new dichromatic reflection model based CSS algorithm was proposed. Experimental results showed that the new algorithm could detect the corner of the objects in image with highlights and shadows correctly, while traditional algorithm obtained many false corners. The new algorithm is better than the traditional algorithms in performance.