Clustering for image retrieval via improved Fuzzy-ART

Clustering technique is essential for fast retrieval in large database. In this paper, new image clustering technique is proposed for content-based image retrieval. Fuzzy-ART mechanism maps high-dimensional input features into the output neuron. Joint HSV histogram and average entropy computed from gray-level co-occurrence matrices in the localized image region is employed as input feature elements. Original Fuzzy-ART suffers unnecessary increase of the number of output neurons when the noise input is presented. Our new Fuzzy-ART mechanism resolves the problem by differently updating the committed node and uncommitted node, and checking the vigilance test again. To show the validity of our algorithm, experiment results on image clustering performance and comparison with original Fuzzy-ART are presented in terms of recall rates.