To obtain access to full text of journal and articles you must register!
- Article name
- Biometric authentication based with thermal facial images based on biometrics to code converters
- Authors
- Lozhnikov P. S., , lozhnikov@gmail.com, Omsk State Technical University, Omsk, Russia
Sulavko A. E., , sulavich@mail.ru, Omsk State Technical University, Omsk, Russia
Zhumazhanova S. S., , samal_shumashanova@mail.ru, Omsk State Technical University, Omsk, Russia
Panfilova I. E., , panfilova_2015@bk.ru, Samara State Technical University, Samara, Russia
Serikova A. E., , , Omsk State Technical University, Omsk, Russia
- Keywords
- multilayer neural networks / biometrics to code neural network converters / automatic learning / facial thermograms / biometric authentication / feature extraction / feature informativeness estimation
- Year
- 2023 Issue 1 Pages 9 - 18
- Code EDN
- OKAIZO
- Code DOI
- 10.52190/2073-2600_2023_1_9
- Abstract
- In the present study, a biometric authentication method was developed based on artificial intelligence neural network models based on thermal facial images in a protected execution mode. "Protected execution" means the impossibility of analyzing the logic of the work of artificial intelligence, controlling artificial intelligence and extracting knowledge from its memory (for example, personal data) by any unauthorized person. The method is based on the artificial neural network InceptionResNet, as well as a modified biometrics-code neural network converter trained according to GOST R 52633.5. The results showed that with this approach, a change in the psychophysiological state of the subject does not lead to a decrease in the accuracy of authentication. The best authentication error rates were: EER = 4.91 (FFR = 0.27 with FAR < 0.001). The proposed method is robust in relation to the user and his state, and is also efficient on small training samples (8 examples of thermograms per person).
- Text
- BUY for read the full text of article
- Buy
- 500.00 rub
