Abstract
Facial recognition is one of the most significant biometric techniques. As a wide research field, it is faced with several challenges; the biggest of which is illumination variations. To overcome this challenge, several various techniques for facial recognition systems have been proposed by researchers. This paper is to improve recognition accuracy when detecting faces in illumination variations. To accomplish that, a face detector to recognize faces under various illuminations was implemented. The proposed model, (H-LBP), utilizes a combination of two phases: preprocessing and features extraction. During preprocessing, homomorphic filtering was used, and during features extraction, local binary patterns was used. The homomorphic filter was used to lower the effect of illumination that comes from an input image. Local binary patterns were used to eliminate illumination variations. The results proved that the combination of different phases was an effective choice since a better accuracy level was achieved than when each of the phases (preprocessing and extraction) was separately used.