WO2006067006A1

WO2006067006A1 - Method and system for detecting biometric characteristics

Info

Publication number: WO2006067006A1
Application number: PCT/EP2005/056069
Authority: WO
Inventors: Jan Lichtermann; Ralph Schmidt; Wolfgang Kluepfel
Original assignee: Robert Bosch Gmbh
Priority date: 2004-12-20
Filing date: 2005-11-18
Publication date: 2006-06-29
Also published as: EP1831818A1; DE102004061335A1

Abstract

The invention relates to a method for detecting biometric characteristics, in particular for identifying living fingerprints. According to the invention, a plethysmogram of the human tissue which is to be tested is produced and the progress of the plethysmogram is compared to predefinable desired values. A deviation in relation to the predefinable desired values indicates the presence of a false object (e.g. a phantom finger). The invention also relates to a system which is used to detect biometric characteristics.

Description

Method and system for acquiring biometric features

State of the art

The invention relates to a method for the detection of biometric features according to the preamble of claim 1. Furthermore, the invention relates to a system for the

Acquisition of biometric features (claims 9 and 10), and an application of plethysmogram analysis in an identity verification system (claims 11 and 12).

For the identification of persons, in particular in connection with

Access control, the acquisition of biometric features is becoming increasingly important. Unlike conventional identification means such as identity cards and the like, such systems and systems are expected to provide greater reliability and security against counterfeiting. As is well known, the fingerprint of a person is a fairly secure tool for one-to-one identification. For this reason, the fingerprint has traditionally been used extensively for identification purposes. It has therefore offered to use the fingerprint in modern systems as Identfϊkationskriterium. Unfortunately, it has been found that these systems are not forgery-proof and therefore pose a potential risk, because they can not exclude the access of unauthorized persons with the required reliability. This problem occurs especially in unmanned monitoring stations, which are easier to manipulate. Since this risk is known, it has already tried automatic

Fingerprint Recognition Systems (AFIS) when used in unsupervised operation equipped with a living finger detection. The live finger recognition should prevent that such systems are overcome by simple finger replicas (fakes). In order to enable a live finger detection, methods have already been proposed which should be suitable for overcoming protection. These include, for example, the measurement of finger temperature, skin impedance, skin resistance, electrocardiogram (ECG), blood oxygen saturation, blood flow, etc.

The listed methods are the measurement of vital parameters of the human body. However, most of the enumerated methods are not necessarily suitable as a live finger detection when examined closely. When measuring temperature, the characteristic "temperature" to be recorded is easy to reproduce. Furthermore, the variance of the measured value is too large to be sensibly detected in a narrow window. The same applies to the measurement of skin impedance and skin resistance. The measurement of the ECG, for example, places very high technical and financial demands on the measuring technology. Furthermore, the measurement of blood oxygen saturation, called pulse oximetry in medical technology, and related methods, such as blood flow measurement, have been proposed, which according to the current state of knowledge should be promising for use in live finger recognition. A common feature of the latter method is the optical non-contact detection of the measured data (vital parameters). The coupling to the human pulse when measuring the blood oxygen saturation has the consequence that at least two pulse cycles must normally be taken into account when detecting the signals. Assuming an average heart rate of 60 beats per minute, a measurement takes an average of two seconds.

Advantages of the invention

The plethysmogram analysis proposed according to the invention measures vital parameters of the human body and decides on the authenticity of the applied finger on the basis of the measurement results. It prevents automatic fingerprint recognition systems with simple technical means and low

Effort (simple fakes) to be overcome. An important advantage of this invention over pulse oximetry is a simplified measurement technique that results in lower hardware costs. Since only light of a certain wavelength is required to record a plethysmogram curve, the second light source used in the Pulse oximetry is needed to be saved. Successful application of the plethysmogram analysis requires less medical knowledge than, for example, pulse oximetry. Clearly definable signals are provided which allow a high degree of certainty to distinguish between the anatomical model of the finger and a technical hose model.

drawing

An embodiment of the invention will be described below with reference to the drawings. It shows

Figure 1 is a block diagram of a system or a flow chart for the living finger recognition;

Figure 2 is a plethysmogram of a living finger;

FIG. 3 shows a plethysmogram of a phantom finder;

FIG. 4 shows a normalized plethysmogram;

FIG. 5 shows a normalized individual pulse of a plethysmogram;

Figure 6 is the block diagram of a system.

Description of the embodiments

The plethysmogram analysis records vital parameters of the human body and, on the basis of the obtained measurement results, decides on the authenticity of a body part, in particular a finger, presented to the measuring system for an examination.

This can be successfully prevented that especially automatic fingerprint recognition systems (AFIS) with simple technical means and little effort (simple fakes) can be overcome. The invention makes use of the fact that when irradiating human tissue, such as a finger, with radiation of a certain wavelength one - A -

characteristic curve, which is called a plethysmogram. The shape of this curve is due to the fact that the absorption of most of the components of the irradiated tissue, such as the skin, connective and fatty tissue, muscles, bones and venous blood, is approximately constant over time. However, this does not apply equally to arterial blood. Namely, at the measurement site, the arterial blood volume is rhythmically increased and decreased due to the heartbeat. As a result of this change in the blood volume, the path length for the radiation used for the measurement also changes, which has an effect on the value of the transmitted intensity. The transmitted intensity varies between two extremes, showing a characteristic of living tissue course. A plethysmogram typical for a living finger is shown in FIG. 2. In this case, the time in seconds is plotted on the x-axis, while the values entered on the y-axis represent normalized amplitude values of the transmitted intensity. Particularly characteristic of a plethysmogram obtained on living tissue are the so-called dicrots. Dicrots are oscillations or delays in the area of the falling edge of a pulse of the plethysmogram, which arise through attenuation in the large capillaries as the blood flows into the extremity being examined. In comparison, Figure 3 shows the plethysmogram of a so-called phantom finger used as a fake with which an AFIS is to be deceived.

FIG. 4 shows a normalized plethysmogram. Scanned values are plotted on the x-axis of the coordinate system and the normalized amplitude of the transmitted intensity on the y-axis. Finally, FIG. 5 shows a normalized pulse of such a normalized plethysmogram. On the x-axis in this case again samples, on the y-axis normalized amplitude values of the transmitted intensity are plotted.

Normalization in this context means that the normalized plethysmogram pulse PN (FIG. 5) has been created by superposing a multiplicity of recorded pulses. Particularly advantageous and easily recognized characteristics of such a pulse are the pulse area, that is to say the area bounded in FIG. 5 by the curve and the x-axis shown there, as well as the time at which the pulse reaches its "peak", that is A particularly accurate examination of a plethysmogram is furthermore possible by additionally determining a correlation coefficient K. For this purpose, the normalized pulse PN shown in FIG. 5 is compared with a reference pulse The correlation coefficient K then gives the degree of correspondence with such a measure Reference pulse on. In an advantageous further embodiment variant of the invention, a twofold transformation of a plethysmogram is compared with a likewise transformed reference signal and in this comparison the quadratic deviation of the comparison partners is determined. The anatomical effects of the

Capillary attenuation detected in the measured extremities by a wavelet transformation and subsequent Fourier transformation. Both error quantities are normalized such that the threshold values for a recognition assume exactly the value one. As a result, a simple application of the method is possible. Furthermore, this also allows a slight adaptation of the predefinable threshold values, which is achieved by the

Use of different measuring sensors may be necessary. In order to achieve the most accurate decision criterion, it has proved expedient, before applying the steps described above, to examine the measurement signal for specific properties which must satisfy specifiable requirements. These properties include, in particular, the mean value, the average power and the defined course of the dicrots already explained above. Particularly advantageous and accurate results can be achieved if the method steps explained above in various embodiments are used in combination. This is explained below with reference to the system diagram shown in FIG. 1, which at the same time reproduces the sequence of such a test.

In a first step 100, measurement data of a test person TP are collected, for example with the system shown in FIG. These measurement data form a characteristic curve, which is called a plethysmogram (FIG. 2). In the following step 101, the measurement data are determined by means of a corresponding

Function module standardized. In step 102, mean value and power are checked, for example by comparing these quantities derived from current measured values with typical values, in particular predefinable limit values. If this comparison does not result in a match, step 102A transfers to step 104, which leads to step 118. In step 118 it is determined that no

Living finger is present, so possibly a deception attempt was made, for example, to outsmart an access control. If it is then determined in step 102 that the values determined from the current measured values correspond to mean value and power of the expected standard, a branch is made to step 102B. This leads to step 103, in which, to increase system security, a further security check takes place. In this verification step, the dicrots in the plethysmogram are examined more closely. Preferably, the shape, in particular the length of the dicrots is analyzed. The dicrots, which are oscillations or delays in the fall of a plethysmogram pulse, are caused by the attenuation in the large capillaries as the blood flows into the limb being examined. They represent particularly characteristic vibrational forms for living tissue. A figure of the dicrots lying outside the expected standard therefore indicates that the current test object is not a living tissue, but a decoy, such as a phantom finger. In this case, turn over the

Steps 103A, 104, step 118 is reached and the determination that no live finger is present. If the data of the measured dicrots are within the permissible range, then the presence of living tissue is assumed and it is branched, via step 103B, to steps 104A and 104B, which lead to different test steps, which will be explained in more detail below. These test steps 104A and 104

B, with each of these branches in subsequent steps, can be carried out independently. Preferably, however, they are used in combination, since this results in increased security against deception maneuvers. On the way via the step 104A, the step 105 is reached, in which a standardized single pulse (FIG. 5) of a plethysmogram is considered. As already described above, in the subsequent steps 106, 107, 108 different criteria of the individual pulse are interrogated and compared with predefinable standard values. Steps 106, 107, 108 may be made individually, but preferably in combination. In step 106, the area underlying the graph of the single pulse is determined. In step 107, a correlation coefficient is determined by comparison with a reference pulse. Finally, in the step

108 determines the time at which the individual pulse reaches its maximum. In the step

109, an assessment is made of the process steps initiated by step 104A. The result of this evaluation is passed via step 114 to a function module which summarizes the evaluation results of the method steps initiated with steps 104A and 104B. However, the branching introduced by step 104B will first be described below. In the subsequent step 110, a wavelet transformation of the measured plethysmogram is formed. In step 111, a corresponding transformation of a reference plethysmogram is provided. Both components Via step 112, a function module is supplied, which compares the two components in step 113, determines the quadratic error deviation and preferably compares this with a desired value. The result is in turn fed via step 114 to a function module which, in step 115, combines the results of the branches initiated with steps 104A and 104B. These

Combination leads to two possible alternatives. Either the step 116 reaches the step 118, in which it is determined that there is no living finger but a delusion object. Or via step 117 step 119 is reached, in which the presence of a living finger is confirmed. If a fingerprint is positively identified while at the same time passing the test for the presence of living tissue, then there is a very high certainty that an authorized person has been recognized. As a result, for example, access can be released as part of an access control. If, on the other hand, a fingerprint can not be identified or the described plethysmogram analysis recognizes a deceptive object, access remains locked. In addition, in the context of the system 6 (Figure 6) a

Warning device can be actuated.

FIG. 6 also shows a schematic block diagram of a system 6. The system 6 comprises a radiation transmitter 60, which preferably emits substantially radiation having a wavelength. Appropriately, here a light emitting diode (LED) can be used, which emits a measuring radiation 64 with the wavelength λ. The system 6 further comprises a radiation receiver 62, for example a sensitive to the radiation of the radiation transmitter 60 photodiode. Reference number 61 designates a test object introduced into the beam path between the radiation transmitter 60 and the radiation receiver 62. This may be, for example, a

Extremity, especially a finger, a test subject act. Reference numeral 63 denotes a functional module that evaluates the output signal of the radiation receiver 62. Furthermore, a function module 65 is provided for the detection of fingerprints. This system 6 thus advantageously enables the detection and analysis of fingerprints and at the same time the determination of whether the finger being tested is a deceiving object or a living finger. This system thus allows a high reliability in the detection of fingerprints with high security against deception maneuvers. In a particularly advantageous development of the invention, the accuracy and reliability of an identity check can be further increased by additionally checking personal characteristics of the plethysmogram apart from the fingerprint of a finger to be checked. This presupposes that, in addition to a library of stored fingerprints, characteristic plethysmograms which are to be assigned to individual persons are also stored. Such an identity check is particularly suitable for access control to security areas to which only a limited number of persons have access, whose fingerprints and plethysmograms are stored there and retrievable for verification purposes.

Claims

claims

1. A method for detecting biometric features, in particular live finger recognition, characterized in that a plethysmogram of the tissue being tested is generated, that the course of the plethysmogram is compared with predefinable setpoint values and that at a predeterminable deviation from the setpoint values on a deception object (eg phantom finger) is closed.

2. The method according to claim 1, characterized in that a normalized plethysmogram is generated by superposition of a plurality of plethysmograms.

3. The method according to any one of the preceding claims, characterized in that measured in the plethysmogram mean and power and compared with predeterminable limits.

4. The method according to any one of the preceding claims, characterized in that the shape of dicrotic plethysmogram, in particular their length is examined and compared with predeterminable limits.

5. The method according to any one of the preceding claims, characterized in that at least one normalized pulse (PN) of a normalized plethysmogram the

Pulse area (PF) is determined and compared with a predetermined setpoint.

6. The method according to any one of the preceding claims, characterized in that at least one normalized pulse (PN) of a normalized plethysmogram of Time (TM) is determined at which the normalized pulse (PN) reaches its maximum (M) and that this time is compared with a predetermined setpoint.

7. The method according to any one of the preceding claims, characterized in that a normalized pulse (PN) with a reference pulse (PR) is compared and from the

Comparison, a correlation coefficient (K) is derived.

8. The method according to any one of the preceding claims, characterized in that a Plethysmogrammkurve is compared with a reference curve, that the square deviation is determined by this comparison and that the square

Deviation is compared with a predetermined setpoint.

9. System (6) for the detection of biometric features characterized by means (60, 62, 63) for performing a plethysmogram analysis.

A system for identity verification, characterized in that the system (6) comprises fingerprint capture and analysis means (65) and means (60, 62, 63) for performing a plethysmogram analysis.

11. Application of plethysmogram analysis for the detection of biometric features in an identity verification system.

12. Application of the Plethysmogrammanalyse according to claim 10 for the Lebendefϊngererkennung.