Hybrid Framework for a Robust Face Recognition System Using EVB_CNN

Tamilselvi M., S. Karthikeyan

Source Title: Journal of Cases on Information Technology (JCIT)23(3)

ISSN: 1548-7717|EISSN: 1548-7725|EISBN13: 9781799859185|DOI: 10.4018/JCIT.20210701.oa4

MLA

Tamilselvi M., and S. Karthikeyan. "Hybrid Framework for a Robust Face Recognition System Using EVB_CNN." JCIT vol.23, no.3 2021: pp.43-57. http://doi.org/10.4018/JCIT.20210701.oa4

APA

Tamilselvi M. & Karthikeyan, S. (2021). Hybrid Framework for a Robust Face Recognition System Using EVB_CNN. Journal of Cases on Information Technology (JCIT), 23(3), 43-57. http://doi.org/10.4018/JCIT.20210701.oa4

Chicago

Tamilselvi M., and S. Karthikeyan. "Hybrid Framework for a Robust Face Recognition System Using EVB_CNN," Journal of Cases on Information Technology (JCIT) 23, no.3: 43-57. http://doi.org/10.4018/JCIT.20210701.oa4

Export Reference

Favorite Full-Issue Download

View Full Text HTML

View Full Text PDF

Abstract

Recognition of the human face is becoming an ingenious technology that enhancing its strategy gradually by finding its applications in a wide variety of fields including security and surveillance. The traditional methods that are in practise for face recognition are not adequate in producing good accuracy due to two main reasons. The first one is the pictures are affected by various uncontrolled situations such as illumination, blur, and pose, and the second one is struggling in an efficient recognition when dealing with a large number of samples. There is need for an effective face recognition as a part of life in the automated environment. The traditional methods are lagging with some parameters. To overcome the aforementioned issues, a new methodology is implemented. This methodology is a hybrid frame work combined with Eigen value-based convolutional neural networks (EVB_CNN). The EVB_CNN is designed in such a way that the significant features are extracted and classified by the softmax function and fully connected layer, respectively. The experimental analysis is carried out with AR data set and ORL data set that shows enhancement in accuracy with significant reduction in computation time with images taken over specific uncontrolled environments.