Contour detection based on Gabor filter and directional DoG filter

Abstract

A contour detector with directional surround suppression, motivated by biological principles, is proposed in this paper. A bank of Gabor filters are used to simulate receptive filed for edge detection. The directional DoG filter is used to simulate the biological surround suppression mechanism. The resulting operator has strong response to the texture boundary, isolated edge and object contour. We use some natural images, with complex textured backgrounds, to test the performance of the proposed model. Result shows that contour detection using this method is more effectively than the classical contour detectors.