Edge-based structural features for content-based image retrieval.pdf

Edge-Based Structural Features for Content-Based Image Retrieval Xiang Sean Zhou, Thomas S. Huang Beckman Institute for Advanced Science and Technology University of Illinois at Urbana Champaign, Urbana, IL 61801, USA {xzhou2, huang}@ifp.uiuc.edu ____________________________________________________________________________________________________________ Abstract This paper proposes structural features for content-based image retrieval (CBIR), especially edge/structure features extracted from edge maps. The feature vector is computed through a ?Water-Filling Algorithm? applied on the edge map of the original image. The purpose of this algorithm is to efficiently extract information embedded in the edges. The new features are more generally applicable than texture or shape features. Experiments show that the new features can catch salient edge/structure information and improve the retrieval performance. Keyword: structural feature; water-filling algorithm; Content based image retrieval; relevance feedback _____________________________________________________________________________________________________________ 1. Introduction Content-based image retrieval (CBIR) is an active yet challenging research area. The performance of a CBIR system is inherently constrained by the features adopted to represent the images in the database. Color, texture, and shape are the most frequently referred ?visual contents? (Flickner et al., 1995). One of the main difficulties in such systems has to do with the fact that the aforementioned ?visual contents?, or low-level features, though extractable by computers, often cannot readily represent the high-level concepts in the user?s mind during the retrieval process. Therefore the research directions include but not limited to, in one direction, incorporating machine learning and intelligence into the system?The learning can be on-line (e.g., user-in-the- loop through relevance feedback) or o