Performance Assessment of Content Based Image Retrieval System using Particle Swarm Optimization Algorithm and Differential Evolution

In this paper, a content-based image retrieval (CBIR) system is presented by employing 12 distance measurements and three types of visual parameters, undergo optimization through particle swarm optimization (PSO) and Differential Evolution (DE) algorithm. Here after, it is called as image retrieval system (IRS) method for the convenience. Initially, IRS derives three types of features of an image: texture, shape and color. Consequently, for every feature type, the similarity among the others and query image in a database D will be estimated, and it uses suitable distance measurements. To optimize the IRS, the closely optimum permutations among the features, similarity metrics and optimum weights for 3 similarities in terms of 3 types of features are determined. In this paper, we made a performance analysis of the application of PSO and DE algorithms to optimize the parameters in the IRS. At the end, simulation outcome shows that the DE method dominates the other traditional methods.