Similarity measures for efficient content-based image retrieval

New similarity measures for comparing two colour histograms are described: the dissimilitude distance DS^* and the similarity distance E. The latter is incorporated into the exponentiation part of the Gibbs distribution and the generalised Dirichlet mixture, while the former is compared to five similarity measures: L₁, L₂ (Euclidean distance), the similarity measure E in addition to Gibbs and Dirichlet distributions integrating E. The proposed measures are implemented into a system called MIRA for an efficient content-based image mining and retrieval. In order to overcome the limitations (and inappropriateness) of some previous information retrieval measures in evaluating the efficiency of an image retrieval process, three variants of a new effectiveness measure are proposed and experimented on an image collection for various similarity measures, including L₁ and L₂. Experimental results show that retrieval effectiveness is the highest for E + Dirichlet and the lowest for the Euclidean distance. They also illustrate the superiority of our approach towards similarity analysis and retrieval effectiveness computation both in the L^* C^* H^* and CIECAM02 colour spaces.