Authors: Jun Zhang, Lei Ye
Addresses: School of Computer Science and Software Engineering, University of Wollongong, Wollongong, Australia. ' School of Computer Science and Software Engineering, University of Wollongong, Wollongong, Australia
Abstract: Feature aggregation is a critical technique in content-based image retrieval (CBIR) that combines multiple feature distances to obtain image dissimilarity. Conventional parallel feature aggregation (PFA) schemes failed to effectively filter out the irrelevant images using individual visual features before ranking images in collection. Series feature aggregation (SFA) is a new scheme that aims to address this problem. This paper investigates three important properties of SFA that are significant for design of systems. They reveal the irrelevance of feature order and the convertibility of SFA and PFA as well as the superior performance of SFA. Furthermore, based on Gaussian kernel density estimator, the authors propose a new method to estimate the visual threshold, which is the key parameter of SFA. Experiments, conducted with IAPR TC-12 benchmark image collection (ImageCLEF2006) that contains over 20,000 photographic images and defined queries, have shown that SFA can outperform conventional PFA schemes.
Keywords: content-based image retrieval; CBIR; feature fusion; series feature aggregation; SFA; threshold estimation; image dissimilarity; visual threshold.
World Review of Science, Technology and Sustainable Development, 2010 Vol.7 No.1/2, pp.100 - 115
Published online: 31 Mar 2010 *Full-text access for editors Access for subscribers Purchase this article Comment on this article