# Method for dynamic biometric personal authentication based on handwriting features

FIELD: physics.

SUBSTANCE: database of users allowed to access a system is formed from sampled and quantised readings of handwriting samples of a matrix of two-dimensional discrete Haar transform coefficients. When allowing a new user into the system, a matrix of two-dimensional discrete Haar transform coefficients similar to the reference sample is formed from the sampled and quantised readings of handwriting samples of that user. Comparison is carried out and a decision is made to associate the obtained record to one of the reference records.

EFFECT: high reliability of authorised access in systems requiring use of personal identification means.

2 cl, 4 dwg

The invention relates to computing, methods of data processing for biometric measurement, in particular to methods of digital processing of handwritten text. The invention can be used in systems restrict access to protected information by means of biometric identification on the individual characteristics of handwriting in addition to existing means of identification or replacement of these facilities.

The progress made in the field of computer data processing [1], has removed many of the technical problems on the implementation of systems identification and monitoring by the motor (motor) mechanisms of man, manifested most fully in the process of writing and speech. Engine dynamic characteristics of the individual shall remain in full individuality in the process of functioning of a living organism, which is what makes them the most informative source of human individuality.

There is a method of identification by the peculiarities of the signature (RU # 2148274 from 27.04.2000, G06K 9/22, 9/62, G06F 15/18), namely, that enter in the computer is converted into digital form fluctuations pen reproducing handwriting identifiable person and its pressure on the tablet, calculate differential and integral parameters of the text, cu is IU, additionally calculates the correlation coefficients of these calculated parameters, and compute the estimates of the correlation coefficients, decide on personal identification by comparing the mentioned parameters calculated from their reference values. The main disadvantage of this technical solution is incomplete use of the received information in the form of the dynamic characteristics of vibration of stylus and pressure, which reduces the reliability of the decision.

The prototype of the claimed invention is a method of biometric authentication according to the handwriting in the computerized access control system (RU # 2469397 C1 from 10.12.2012, G06K 9/00) consists in the fact that multi-component analog signal is converted into digital form, forming the matrix of quantized samples, the values of its elements is calculated using two-dimensional discrete cosine transform matrix coefficients, the elements of which are used as identification parameters of the user, any user that logs access control, offer to perform in a similar way the pattern writing handwriting handwriting received multi-channel analog signal is converted into digital form, calculated using two-dimensional discrete cosine transform Mat the Itza coefficients, elements which are compared with the corresponding elements of the matrices of coefficients of two-dimensional discrete cosine transform of registered users in the database, which is recognized by the user is considered incidental reference recording, if the difference is minimal.

The main disadvantage of this method is that the Fourier transform, in particular two-dimensional discrete cosine transform is ill-suited to work with functions defined on a short interval argument, and functions whose value dramatically and significantly changed.

The analysis of the level of technology has allowed to establish that the analogues, characterized by a set of characteristics that are identical to all features of the claimed technical solution is available, which indicates compliance of the claimed method the condition of patentability "novelty".

The aim of the invention is to develop a method of dynamic biometric identity authentication on the peculiarities of the handwriting handwriting, allowing to increase the reliability of the verification of the user.

This objective is achieved in that the determination of the individual characteristics of handwriting produced by the identity matrix, obtained by two-dimensional discrete Haar transformation, this creates a base et the h-parameters of the samples of the handwriting of the user, made in the computerized system, in the form of interval changes each identification parameter, newly introduced to the system user identity is an identity matrix, obtained by two-dimensional discrete Haar transformation, thus calculate the maximum and minimum values of the identifiable parameters, and the decision on the admission of any user in the system is accepted, if the Hamming distance is minimum, which determine the total number of depositions of the identification parameters for the intervals of admissible values of the reference parameters of the samples of the handwriting of users admitted in the computerised system, otherwise form the message about unauthorized attempt logon, the intervals of admissible values of the reference parameters of the samples of the handwriting of users admitted in the computerised system, artificially increased by 20-40% compared with calculated values.

This set of essential features allows you to create a computerized system for the decision support using user authentication on the features of the handwritten text.

The implementation of the method is illustrated by drawings, where Fig.1 presents a conceptual model of the proposed method, Fig.2 - p is intitially scheme of processing of multidimensional analog signal, in Fig.3 shows an example of forming a matrix of two-dimensional discrete Haar transformation, Fig.4 - forming vector quantized samples from a single channel measurements.

The proposed method consists of three consecutive stages: the stage of creating a database of reference parameters of the handwriting of the users who have access to the computerized system, stage recording and processing parameters handwriting handwriting of any user included in the computerised system, and the stage of the decision on admission. These stages are shown in Fig.1.

The first stage envisages the creation of a database of the identification parameters of the samples of the handwriting of users admitted in the computerised system, this involves the following steps. Using a device for measuring motion parameters of stylus, such a device is described in patent RU №2475699 C2 from 20.02.2013, G01B 7/00), issued in the name of the applicant, it is recorded handwriting sample (Fig.2). This patent uses the maximum amount of initial information, analyzing the dynamics of writing handwriting (the trajectory of the writing unit for writing text - curves fluctuations writing site for axis Y(t), X(t), varying the pressure of writing unit for writing text - pressure curve P(t).

The multichannel processing the tax signal is carried out in accordance with patent No. 2469397, G06K 9/00 from 10.12.2012, issued in the name of the applicant. On the basis of a set of quantized discrete samples multichannel signal form the matrix of quantized samples (Fig.3, 4), bearing original information about the features of the writing of the text. For the formation of an identity matrix with the allocation of identification parameters spend two-dimensional discrete Haar transform, described, for example, in the book [2]. Thanks to a more complete representation of the local features of multidimensional signal, missing the Fourier series, wavelets, in particular, the Haar transform, have found practical application for analysis of subtle features of multidimensional signals. Two-dimensional discrete Haar transform includes the processing of the matrix is N×N values, where the first operations are performed with each row of the matrix, and then the same operation is performed for each column of the result. The conversion is performed according to the following method. For example, consider the matrix 4×4:

The first step of the two-dimensional discrete Haar transformation consists of a one-dimensional transform of each row by multiplying the matrix PW. Here we use the matrix W is orthonormal function Haar fourth order [2]: