Image matching is a fundamental task of many problems in computer vision. This paper presents a novel local feature descriptor based on the gradient distance and orientation histo...
This paper presents a new keypoint-based approach to nearduplicate images detection. It consists of three steps. Firstly, the keypoints of images are extracted and then matched. S...
Abstract. In this paper, we present novel image-derived, invariant features that accurately capture both the geometric and color properties of an imaged object. These features can ...
Sign language recognition constitutes a challenging field of research in computer vision. Common problems like overlap, ambiguities, and minimal pairs occur frequently and require...
This paper presents a novel method for image classification. It differs from previous approaches by computing image similarity based on region matching. Firstly, the images to be ...