A Novel Approach for Interest Point Detection Based on Phase Congruency

In the paper, a novel method of interest points detection based on phase congruency by using Log-Gabor wavelets is proposed. Points of maximal phase congruency in phase domain correspond to all the different types of 2D features in an image, including gray-level corners, line terminations, and a variety of junctions within lines. Phase congruency is a dimensionless measure invariant to the changes of image brightness and contrast; hence it allows the use of universal threshold values that can be applied over wide classes of images. Moreover, different kinds of interest points can be detected since it does not make any assumption on the geometric structure of interest points in the proposed method. Experimental results on synthetic and natural scene images have shown that accuracy of detection and localization of interest points can be achieved using the proposed approach.