Bridging the Semantic Gap for Texture-based Image Retrieval and Navigation

Abstract

In this study, we propose a new semantic approach for interpreting textures in natural terms. In our system, the user can reach desired textures by navigating into a hierarchy of sub collections previously held (offline). The originality of the proposed approach stems from two reasons: (1)- the intrinsic properties of the texture features extracted from the co-occurrence matrices have never been used before and (2)- it provides some degree of tolerance to generate the classes semantic which is not available with the standard unsupervised clustering algorithms such as kmeans. Thus, our contibutions in this study are threefold. (1)- Our approach maps low-level visual statistical features to high-level semantic concepts; it bridges the gap between the two levels enabling to retrieve and browse image collections by their high-level semantic concepts. (2)- Our system models the human perception subjectivity with the degree of tolerance and (3)- it provides an easy interface for navigating and browsing image collections to reach target collections. A comparative study with the unsupervised clustering algorithm k-means reveals the effectiveness of the proposed approach.