Smart Attendance System Using Multi-Classifier Face Recognition

Abstract

Taking attendance in an Institution of learning or an organization cannot be over emphasized. However, keeping attendance records manually can be tedious and ineffective due to time wastage. Researchers have come up with different automated systems of attendance like the Radio Frequency Identification technology (RFID) and Barcode technology to address the issue of time wastage, ease of use and convenience. However, these systems are susceptible to spoofing. An automated system that has the mechanism to proof a claimed identity and at the same time be convenient to accurately process attendance is crucial.

This study is developed through the fusion of two feature extraction algorithms: Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), aiming at increasing the recognition accuracy compared to using either PCA or LDA singly. The system design follows the divide and conquer method and was implemented using MATLAB R2015a. The recognition performance rate of the system objectively justifies the fusion of PCA and LDA thereby improving the objects recognition accuracy compared to using either PCA or LDA in face recognition. With the appropriate alarm threshold values the identification rate of 60%, 40% and 60%; false negative identification error rate of 0%, 0% and 0%; and the false alarm rate of 40%, 60%, and 40% were recorded.