Development and Simulation Analysis of a Robust Face Recognition Based Smart Locking System

Face recognition based smart locking systems are susceptible to variation in ambient light conditions. This paper presents a robust face recognition based smart locking system. The novelty of our work is that the choice of the algorithm for face detection and recognition is based on the intensity of light at that time. This system uses basic principal component analysis, linear discriminant analysis, and its variants for face detection and recognition. Access is granted to the user if their image matches one in a predefined database. In case the light intensity is so low that no algorithm gives a satisfactory result, our system will authenticate via a Bluetooth-based one-time passcode. MATLAB/Simulink was used to simulate this system which was subsequently prototyped. The developed prototype had 90% accuracy in low light conditions when larger training databases are used and 90% accuracy in normal light conditions when smaller training databases are used.