METHOD OF EXPRESS-TESTING MEANS OF HIGHLY RELIABLE BIOMETRIC-NEURAL NETWORK AUTHENTICATION OF A PERSON USING A "FRIEND" BIOMETRIC IMAGES BASE Russian patent published in 2020 - IPC G06K9/62 

FIELD: computer equipment.

SUBSTANCE: invention relates to computer engineering for biometric identification and personal authentication. In the declared method, the biometric-code converter is trained on images of the "Friend" base in order to obtain "Foe" images, carrying out morphing and mutation operations of examples of the image "Friend", creating a base of synthetic examples of the image "Foe", from which pairs of images-parents "Foe" are formed, parallel performance of mutation and morphing procedures of "Foe" parent image pairs in order to preserve correlation matrix of "Foe" images, permutations of mathematical expectations of biometric parameters of images, cyclic supply of one synthetic example of image "Foe" to input of trained neural network converter of biometry-code and after reception of access key code at its output, its removal from memory of "Friend" base, comparison of output codes of access key "Friend" and "Foe" with Hemming measure, calculation of error probability of 2nd kind.

EFFECT: technical result consists in reducing the volume of the "Foe" image base and the computational resources required for its testing while maintaining the reliability of testing high-reliability biometric-neural network authentication of the individual on the probability of occurrence of errors of the second kind.

1 cl, 2 dwg