P. V. Kozlov - Post-graduate Student, Perm National Research Polytechnic University E-mail: vfrey@mail.ru А. А. Yuzhakov - Dr.Sc. (Eng.), Professor, Department of Automation and Telemechanics, Perm National Research Polytechnic University E-mail: uz@at.pstu.ru

Pattern Recognition at the moment is difficult, is not fully explored and demand issue. There are many algorithms for the recognition of various objects, but they are all approximate. The authors consider the recognition method based on the representation of the human brain thinking. Exemplify neuronal network based on cognitron. This network is capable of self-organization as a result, that is, learning without a teacher to learn a variety of input images and react to their appearance in the future of their badges. Conducted testing of the network, taking into account the presence of noise in the input signal.

 

1. Kozlov P.V., Lipin JU.N., JUzhakov A.A. Raspoznavanie lica cheloveka. Informacionnye tekhnologii v nauke, obrazovanii, telekommunikacii  biznese (IT + SE-11) // Materialy XXXVIII Mezhdunar. konf. i diskus. nauch. kluba (Ukraina, Krym, JAlta-Gurzuf, 1-10 oktjabrja 2011). Prilozhenie k zhurnalu «Otkrytoe obrazovanie». 2011.
2. Uossermen F.Nejjrokompjuternaja tekhnika: Teorija i praktika. NeuralComputing. Theory and Practice. M.: Mir. 1992. 240 s.
3. Fukushima K. Cognitron: A self-organizing mult ilayered neural network // Biological Cybernetics. 1975. V. 20. P. 121-36.
4. Fukushima K. Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition uneffected by shift in position // Biological Cybernetics. 1980. V. 36(4). P. 193-202.