Human Face Recognition using Eigen Vectors, MATLAB, and AI

Abstract

Real-world face recognition system’s applications come with problems with lighting, pose, and facial expressions. This research work proposes the integration of Principal Component Analysis (PCA) for dimensionality reduction and Deep Neural Networks (DNN) for classification in order to enhance both accuracy and speed. On the Yale and ORL Face Databases, the system was able to achieve higher recognition rates; DNN was able to recognize faces with 97.1% on the Yale and 95.4% on the ORL as compared to Support Vector Machines. The integration of PCA helped to reduce computational load and therefore made the system more suitable for applications in real-time such as security and surveillance and biometric authentication. The work enhances the model’s resistance to lighting, pose, and expressions changes, which are critical issues in current systems. Future work will look into how occlusions and the extremes of pose can be dealt with.