SYNTHESIS OF TEST IMAGES FOR STABILITY ESTIMATION OF NETWORK TRANSFORMERS IN BIOMETRIC IDENTIFICATION SYSTEMS 

Grishko Aleksey Konstantinovich, candidate of technical sciences, associate professor, sub-department of radio equipment design and production, Penza State University (440026, 40 Krasnaya street, Penza, Russia), alexey-grishko@rambler.ru 
Lukin Vitaliy Sergeevich, postgraduate student, Penza State University (440026, 40 Krasnaya street, Penza, Russia), Kipra@pnzgu.ru 
Yurkov Nikolay Kondrat'evich, doctor of technical sciences, professor, head of sub-department of radio equipment design and production, Penza State University (440026, 40 Krasnaya street, Penza, Russia), yurkov_NK@mail.ru

004.056; 51-76

10.21685/2307-4205-2017-2-5

Background. Modern society assumes active use of Internet resources. The current practice of password protecting access to personal offices has a significant vulnerability. To strengthen the protection of access to electronic resources of state agencies and offices to personal users the necessary technology for biometric authentication of the individual by the transformation of personal biometric data of a person in his cryptographic key or a long password. To achieve this goal using special converters based on neural networks. In this case, information security depends on the strength of the neural network converters and generating the procedure of its evaluation test of biometric images.The purpose is to develop biometric handwritten test images for evaluating the resistance of neural network converters biometrics code when using the program.
Materials and methods. The article uses methods of simulation and theory of neural networks. Used simulation environment and the Mathcad environment Bioneurological.
Results. The synthesized model test biometric handwritten images. A comparison of statistics obtained from the source images. The model is tested in the environment of Bioneurological. 
Conclusions. Using this model it is possible to assess the resistance of neural network converters biometrics code. The model allows to use a limited source of information and significantly reduce the time to create images. This will lead to amenesia temporary, physical and economic costs of the assessment of persistence of neural network converters.

neural network, biometrics, biometric image, durability, model