SCALE-BASED DETECTION OF CORNERS OF PLANAR CURVES

A technique for detecting and localizing corners of planar curves is p roposed. The technique is based on Gaussian scale space, which consist s of the maxima of absolute curvature of the boundary function present ed at all scales. The scale space of isolated simple and double corner s is first analyzed to investigate the behavior of scale space due to smoothing and interactions between two adjacent corners. The analysis shows that the resulting scale space contains line patterns that eithe r persist, terminate, or merge with a neighboring line. Next, the scal e space is transformed into a tree that provides simple but concise re presentation of corners at multiple scales. Finally, a multiple-scale corner detection scheme is developed using a coarse-to-fine tree parsi ng technique. The parsing scheme is based on a stability criterion tha t states that the presence of a corner must concur with a curvature ma ximum observable at a majority of scales. Experiments were performed t o show that the scale space corner detector is reliable for objects wi th multiple-size features and noisy boundaries and compares favorably with other corner detectors tested.