Affine invariant local features based on novel keypoint detection and grouping

Feature detection is an important topic in computer vision and image processing. In this paper we propose a fixed-iteration algorithm to detect keypoints based on a ridge-type image. We also present a method to group all keypoints on the same edge to form a new affine invariant feature for recognition in images. Experimental results show that the proposed method finds keypoints more exactly and stably than other methods and has good repeatability.