There have been important recent advances in object recognition through the matching of invariant local image features. However, the existing approaches are based on matching to i...
Recent research in object recognition has demonstrated the advantages of representing objects and scenes through localized patterns such as small image templates. In this paper we...
This paper presents a semi-supervised learning (SSL) approach to find similarities of images using statistics of local matches. SSL algorithms are well known for leveraging a larg...
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 ...
A ubiquitous environment enable us to enjoy various services “anytime” ”anywhere”. However, “everyone” is not realized. We research an intelligent space “everyone” ...