A Region-based Approach to Conceptual Image Classification

Classifying images into a set of semantic categories that are meaningful to humans has proved to be a challenging and attractive problem in the field of content-based retrieval. Addressing this problem is typically based on the initial extraction of low-level features for the images and the subsequent application of a pattern recognition technique, to divide the feature space in a number of subspaces corresponding to the semantic categories. An extension to this framework is presented in this paper, aiming at the improvement of the efficiency of image classification systems. This is based on the introduction of an unsupervised still image segmentation algorithm to the process and its combination with MPEG-7 low-level descriptors and a Bayes classifier. Experimental results using different pairs of classes and corresponding data sets demonstrate the efficiency of the proposed approach.