An Efficient Face Recognition System Based on PCA and Extended Biogeography-Based Optimization Technique

The current work is primarily focused on face recognition where the results are calculated on accuracy of face prediction based on a hybrid algorithm which is about biogeography-based optimization (BBO) and extended biogeography-based optimization taken in series with principal component analysis (PCA) to analyze the performance of the proposed model. Four primary parameters are taken for face recognition namely false acceptance, false rejection, processing time and recognition accuracy. To make a fair comparison of techniques discussed in this work, the results of three different techniques discussed in the past have been compared with the proposed work. The complete work has a prime identity claiming method in which the work proposed in the form of BBO and extended BBO has outperformed by making a tradeoff between processing time and recognition accuracy. The percentage accuracy comparison has been made between the proposed work along with particle swarm optimization (PSO) and Exhaustive Search Technique (EST). Though, the exhaustive search technique is good but BBO offers smooth ROC when compared to other techniques. Experimental results show that the BBO-based feature selection algorithm is found to generate great recognition results with 72% accuracy and even better with 90.25% accuracy of extended BBO with processing time of 0.00249 s and 0.00453 s, respectively, which shows a significant improvement in the proposed method.