Performance Evaluation of Hybrid Swarm Intelligence Techniques for Extraction and Recognition of Facial Features

Abstract

The two most important steps in the face recognition procedure are Feature Extraction and Feature Selection. Metaheuristic approach is applied for recognition of a face in this work. First of all, the features are selected using the Discrete Cosine Transformation technique. On these selected features Particle Swarm Optimization technique is applied to select the meaningful features. In the end, Ant Colony Optimization approach is applied on those selected features to recognize the facial image. This is done by measuring distance between the features selected. The results are compared using the parameters Accuracy, Training Time, Recognition Time, Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR). The results show the significant increase in the accuracy, PSNR and MSE values using hybrid approach than the Ant Colony Optimization and Particle Swarm Optimization techniques.