University Links: Home Page | Site Map
Covenant University Repository

Development of an Illumination Invariant Face Recognition System

Okokpujie, Kennedy O. and John, S. N. and Ufuah, Donald and Nwagu, Martins and Noma-Osaghae, Etinosa and Ndujiuba, Charles Uzoanya and Okokpujie, I. P. (2020) Development of an Illumination Invariant Face Recognition System. International Journal of Advanced Trends in Computer Science and Engineering, 9 (5). pp. 9215-9220. ISSN 2278-3091

[img] PDF - Published Version
Download (374kB)
[img] PDF
Download (374kB)


The Face recognition systems have gained much attention for applications in surveillance, access control, forensics, border control. Face recognition systems encounter challenges due to variation in illumination, pose, expression, occlusion and most importantly, aging. The effect of the intensity of light on recognition image in contract with gallery image significantly affect the face recognition system. In this study, an illumination invariant Face Recognition System is developed using a 4-layered Convolutional Neural Network (CNN). The proposed system was able to recognize the different degree of face Illuminated image, thus making the model Illumination Invariant Face Recognition system. The variations caused by illumination was modelled as a form of light varying noise, and it was validated by computing its error statistics and comparing its performance with existing models found in literature. The result of the study showed that an adaptive and robust face recognition system that is illumination invariant could be achieved with CNN. The recognition accuracy achieved by the study was 99.22% with five (5) epochs and iteration of 85.

Item Type: Article
Subjects: T Technology > T Technology (General)
T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions: Faculty of Engineering, Science and Mathematics > School of Civil Engineering and the Environment
Depositing User: Engr. Kennedy O. Okokpujie
Date Deposited: 02 Nov 2020 09:52
Last Modified: 02 Nov 2020 09:52

Actions (login required)

View Item View Item