Corner Detection Using Second-Order Generalized Gaussian Directional Derivative Representations

Corner detection is a critical component of many image analysis and understanding tasks, such as object recognition matching. Our research indicates that existing corner algorithms cannot properly depict the difference between edges corners this results in wrong detections. In paper, capability second-order generalized (isotropic anisotropic) Gaussian directional derivative filters to suppress noise evaluated. The representations step edge, L-type corner, Y- or T-type X-type star-type are investigated obtained. A number properties for discovered which enable us propose new method. Finally, criteria on accuracy average repeatability under affine transformation, JPEG compression, degradation, region used evaluate proposed detector against nine state-of-the-art methods. experimental show our outperforms all other tested detectors.