Authentication System Using 3D Face With Algorithm DLT and Neural Network

Pane, Ivransa Zuhdi and Alexander, Leonardus and Kusnadi, Adhi and Wella, Wella and Winantyo, Rangga (2018) Authentication System Using 3D Face With Algorithm DLT and Neural Network. 2018 Joint 10th International Conference on Soft Computing and Intelligent Systems (SCIS) and 19th International Symposium on Advanced Intelligent Systems (ISIS).

Full text not available from this repository.

Abstract

Many computer systems and data storage are accessed by unauthorized parties, due to the lack of authentication quality possessed by the system. Face Recognition is one of the evolving authentication systems at this time, which can use 2D or 3D data. However, 2D face recognition is easily influenced by environmental circumstances, facial orientation, facial expressions, and makeup. The use of 3D data can help to overcome the intrinsic problems possessed by the 2D approach. 3D face recognition system using a 3D camera has a weakness, i.e. it cannot be used in outdoor. We proposed the usage of DSLR commercial cameras to overcome such problem. The algorithms used in this research is are DLT and neural network. DLT is an algorithm that can determine the 3D coordinates of a point obtained from some 2D images. The neural network is used to recognize faces. In this study, an authentication system is made using DLT and neural network algorithms using DSLR cameras for capturing the faces. The best combination of neural network architecture is the number of hidden nodes 20 with 1 layer, learning rate 0.005 with an accuracy of 95%, the percentage of FAR and FRR of 5% and 5%.

Item Type: Article
Subjects: 000 Computer Science, Information and General Works > 000 Computer Science, Knowledge and Systems > 004 Computer Science, Data Processing, Hardware
000 Computer Science, Information and General Works > 000 Computer Science, Knowledge and Systems > 005 Computer Programming
Divisions: Faculty of Engineering & Informatics > Informatics
Depositing User: mr admin umn
Date Deposited: 05 Oct 2021 03:53
Last Modified: 05 Oct 2021 03:53
URI: https://kc.umn.ac.id/id/eprint/18495

Actions (login required)

View Item View Item