Time Attendance Using FELE Face Identification Algorithms

Abstract

Time attendance system has been used to check-in and check-out of employees in organizations for many years. Biometric technique such as the fingerprint scanner is among the popular and widely used in the past and present. Nowadays, face recognition techniques have been employed to identify the faces such as eigen face, fisher face and local binary patterns histograms (LBPH). These techniques provide acceptable results but may not accurate enough due to a number of false positive cases. Therefore, this paper proposes a better face identification technique to reduce the false positive cases. The method presents the combination of the above three existing algorithms to produce the more accurate result for time attendance by means of the actual experiment. The experiment is carried out using the 30 employees face each with 50 images with a total of 1,500 images as a face database. The testing process contains real faces of 20 employees and 5 non-employees. These faces are tested against the Eigenface, Fisherface, LBPH and the proposed method Fisherface, Eigenface, LBPH Extension (FELE) algorithms. We present and compare the results among these four techniques in term of the false positive, accuracy and precision. It has been shown that the FELE has an accuracy of 100% and outperforms other methods in all categories.