Binary image matching using scale invariant feature and hough transforms

Abstract

Scale Invariant Feature Transform (SIFT) is used for local features description of images. The proposed technique employs SIFT in order to match binary images. While the employment of SIFT descriptors on binary image databases could be possible, its power is rather limited. The novelty of the proposed algorithm is make use of SIFT for binary image matching by taking the power of Hough Transform (HT) in line detection. HT can be used on any line orientation. Thus, HT investigates lines as criteria for binary image similarity beside SIFT features for local and corner descriptions. The evaluation is achieved and experimental highlights the superiority of this approach for binary images which contains straight lines.