An integrated approach to region based image retrieval using firefly algorithm and support vector machine

Intelligent image retrieval is a challenging technology in multimedia applications where bridging the gap between user׳s expectation and low level features is typically hard for computing systems. In the proposed approach, a unique method is projected which integrates support vector machine based learning with an evolutionary stochastic algorithm, called firefly algorithm as a relevance feedback approach into a region based image retrieval system. This system overcomes the semantic gap through optimized iterative learning and also provides a better exploration of solution space. Support vector machine learning automatically updates the weights of preferences for relevant images based on the both relevant and irrelevant feedback images. The firefly optimizer guides the swarm agents to move towards the cluster of relevant images in the exploration of the search space based on user׳s feedback. This research study has a focused approach to increase the performance by optimizing region feature with the firefly algorithm. The efficiency of the proposed approach is experimented on the standard subset of Corel, Caltech and Pascal database images. The performance of the proposed approach is compared with other existing retrieval methods like particle swarm optimization, genetic algorithm, support vector machine and query point movement to identify the excellence with regard to the model in terms of precision and recall. •Support vector machine and Firefly algorithm is incorporated into region based image retrieval as relevance feedback approach inorder to reduce semantic gap.•Support vector machine learns from user relevance feedback and the firefly algorithm not only serves as an efficient optimization engine, but also explores the solution space effectively.•The proposed approach yields higher average precision and recall when compared to existing algorithms.