IMAGE PREPROCESSING WITH SYMMETRICAL FACE IMAGES IN FACE RECOGNITION AND REGRESSION CLASSIFICATION

Abstract

Face recognition with variations in lighting is a significant thing than the physical
characteristics of each individual in facial recognition problems. The difference in illumination from the right
and left can affect the face image. A face image is an axis-symmetrical object. There are many studies on
face recognition, but only a little attention is paid to this issue and few studies to explore and exploit the axissymmetrical property of faces for face recognition are conducted. In this paper, we take the axis-symmetrical
nature of faces then take the side of the face that is not exposed to dark light for pre-processing that will then
be done by robust regression classification. The results of this experiment got an average accuracy of 92.8%.
Using the proposed method on data with a 51 to 77- degree illumination angle produces better accuracy.