A novel content-based image retrieval system based on Bayesian logistic regression

Abstract

In this work, a novel content-based image retrieval (CBIR) method is presented. It has been implemented and run on “Qatris IManager” [14], a system belonging to SICUBO S.L. (spin-off from University of Extremadura, Spain). The system offers some innovative visual content search tools for image retrieval from databases. It searches, manages and classifies images using four kinds of features: colour, texture, shape and user description. In a typical CBIR system, query results are a set of images sorted by feature similarities with respect to the query. However, images with high feature similarities to the query may be very different from the query in terms of semantics. This discrepancy between low-level features and high-level concepts is known as the semantic gap. The search method presented here, is a novel supervised image retrieval method, based in Bayesian Logistic Regression, which uses the information from the characteristics extracted from the images and from the user’s opinion who sets up the search. The procedure of search and learning is based on a statistical method of aggregation of preferences given by Arias-Nicolás et al. [1] and is useful in problems with both a large number of characteristics and few images. The method could be specially helpful for those professionals who have to make a decision based in images, such as doctors to determine the diagnosis of patients, meteorologists, traffic police to detect license plate, etc.