Logistic regression models for a fast CBIR method based on feature selection

Distance measures like the Euclidean distance have been the most widely used to measure similarities between feature vectors in the content-based image retrieval (CBIR) systems. However, in these similarity measures no assumption is made about the probability distributions and the local relevances of the feature vectors. Therefore, irrelevant features might hurt retrieval performance. Probabilistic approaches have proven to be an effective solution to this CBIR problem. In this paper, we use a Bayesian logistic regression model, in order to compute the weights of a pseudo-metric to improve its discriminatory capacity and then to increase image retrieval accuracy. The pseudo-metric weights were adjusted by the classical logistic regression model in [Ksantini et al., 2006]. The Bayesian logistic regression model was shown to be a significantly better tool than the classical logistic regression one to improve the retrieval performance. The retrieval method is fast and is based on feature selection. Experimental results are reported on the Zubud and WANG color image databases proposed by [Deselaers et al., 2004].