Automated face retrieval using bag-of-features and sigmoidal grey wolf optimization

Abstract

A content based image retrieval system is one of the prime research fields due to the exponential increasing of multimedia data over Internet especially images. Although, a number of content based image retrieval methods have been introduced, it is still a challenging task specially for face recognition. Therefore, this work presents an automated face retrieval system using an enhanced bag-of-features framework. The bag-of-features framework has been modified by incorporating a new sigmoidal grey wolf optimization algorithm. The sigmoidal grey wolf optimization algorithm uses a sigmoid decreasing function to escape it from local optima. The efficiency of the proposed sigmoidal grey wolf optimization algorithm has been analyzed over various standard benchmark functions for average fitness values and convergence behavior. Furthermore, it has successfully been used to generate the codewords in bag-of-features framework. The modified bag-of-features has been utilized in content based image retrieval for Oracle Research Laboratory (ORL) face database. The simulation results represent that the proposed method effectively retrieves the faces as compared to other nature-inspired based methods.