JP7168259B2 - living body detector - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Law 1. Announced in publication (1) ・Publication name: Proceedings of the General Conference 2017 (DVD Proceedings) ・Publisher: The Institute of Electronics, Information and Communication Engineers ・Publication date: March 7, 2017

Application of Article 30, Paragraph 2 of the Patent Law 2. Presentation at meeting (1) ・Meeting name: 2017 IEICE General Conference ・Venue: Meijo University Tenpaku Campus (Nagoya City, Aichi Prefecture) ・Date: March 23, 2017

Application of Article 30, Paragraph 2 of the Patent Law 3. Announced in publication (2) ・Publication name: The Institute of Electronics, Information and Communication Engineers Technical Research Report (IEICE Technical Report), vol. 117, No. 2, AP2017-11, pp. 55-58 ・Published by The Institute of Electronics, Information and Communication Engineers ・Published on April 13, 2017

Application of Article 30, Paragraph 2 of the Patent Law 4. Presentation at meeting (2) ・Meeting name: The Institute of Electronics, Information and Communication Engineers, Antenna and Propagation Study Group (AP) ・Venue: Toyonaka Campus, Osaka University (Toyonaka City, Osaka Prefecture) ・Date: April 20, 2017

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a living body detection device that determines whether an object to be detected is a living body.

Biometric authentication techniques such as fingerprint authentication and vein authentication use information that only the person himself/herself can have, so they are more convenient and secure than conventional authentication using IC cards or passwords. However, since there is concern about "spoofing" damage using physical camouflage that simulates the biometric information of another person, in addition to biometric authentication technology, biometric detection technology to distinguish between biometrics and camouflage will be used. This is very important.

Accordingly, one of the inventors of the present invention proposed a BPF (band-pass filter) in which two sets of CSRR (Complementary Split-Ring Resonator) structures are arranged in series on the ground plane of a microstrip line. Pass Filter: band-pass filter) (CSRR-BPF) is used as a living body detection sensor, and the electromagnetic wave response characteristics are measured when the detection target part of the subject is brought close to or in contact with the CSRR-BPF, and the measured value and We have already reported a living body detection device that determines whether a detection target part is a living body or a camouflage by using two evaluation indices, average difference and similarity calculated from living body reference values (Patent Document 1).

JP 2015-211798

The living body detecting device described above was able to determine whether the living body was a living body or a camouflaged object with a certain degree of accuracy. However, it is not possible to completely eliminate false detections for thin camouflage objects that have characteristics close to the human body, such as skin phantoms that simulate the electrical constants of the skin on the surface of a human finger. There was a need for improved accuracy.

As a factor of erroneous detection, the CSRR-BPF used as a living body detection sensor in the living body detection device described above has a structure in which two sets of CSRR structures are arranged in series. One of the reasons for this is thought to be that when the attached finger covers the CSRR portion, which is the area where the CSRR-BPF is placed, variations occur due to imperfect adhesion. Therefore, if the mounting area of the detection target part of the living body detection sensor can be reduced, it is possible to stably ensure the close contact between the detection target part and the living body detection sensor, leading to a reduction in false detection. There has been a demand for a more compact living body detection sensor.

On the other hand, it was also necessary to reconsider the two evaluation indices, average difference and similarity, used to discriminate between a living body and a camouflaged object in the living body detecting device described above. Even with the same detection target part of the same sample, the measured values of the electromagnetic wave response characteristics have a certain amount of variance due to the variations in the CSRR section as described above, but the degree of similarity is an evaluation index that does not allow for this variance. It was thought that there was a factor in the above-mentioned false detection. Therefore, instead of the degree of similarity, a new method that allows the dispersion of the electromagnetic wave response characteristics when there is no camouflage attached and emphasizes and extracts the change in the electromagnetic wave response characteristics when the camouflage is attached It was necessary to introduce an evaluation index. In addition, it has been required to improve the average difference so as to improve the discrimination accuracy.

The present invention has been made in view of the above circumstances. It is an object of the present invention to provide a living body detecting device with improved discrimination accuracy compared to the conventional one by introducing an index.

A living body detection apparatus according to the present invention includes a band-pass filter having a split ring resonator designed to have a predetermined center frequency according to a detection target site of a subject, and generating an electromagnetic wave to pass the detection target site through the band-pass filter. A measurement unit for measuring electromagnetic wave response characteristics in a predetermined frequency band when brought close to or in contact with a filter, a measured value of the electromagnetic wave response characteristics measured by the measurement unit, and a reference value of the electromagnetic wave response characteristics of a living body. and an evaluation value calculation unit that calculates an evaluation value using wherein the evaluation value calculator calculates the measured value of the electromagnetic wave response characteristic measured by the measuring unit as S(f), the reference value of the electromagnetic wave response characteristic of the living body as T(f), Let σ(f) be the standard deviation of T(f), k(f) be the consistency index represented by the following equation (2), and w _v (k(f)) be the weighting function corresponding to k(f). , the weight function for each frequency is w _f (f), the lower limit of the frequency band used for evaluation is f _L , and the upper limit is f _H. It is characterized by calculating the evaluation value α.

Further, in the living body detection device according to the present invention, the weight function w _v (k(f)) is a function represented by the following equation (3), where p is a predetermined threshold value. Characterized by

Further, in the living body detecting device according to the present invention, the evaluation value calculation unit uses a second evaluation value β represented by the following equation (4) in addition to the first evaluation value α as the evaluation value. is characterized by calculating

According to the living body detection device according to the present invention, since the CSRR-BPF has a structure in which two sets of CSRRs are arranged coaxially, compared with a conventional structure in which two sets of CSRRs are arranged in series, the object to be detected is Since the placement area for covering the CSRR part can be reduced depending on the part, variation in measuring the electromagnetic wave response characteristics is reduced, thereby improving the accuracy of determining whether the part to be detected is a living body or not.

According to the living body detection device according to the present invention, the evaluation value calculation unit calculates the first evaluation value α as an evaluation value. This is an evaluation value that allows variations in the measured values of the characteristics, while emphasizing and extracting changes in the measured values of the electromagnetic wave response characteristics when a camouflage is attached. The accuracy of determining whether or not is improved.

Furthermore, by appropriately setting the weighting function w _v (k(f)) used to calculate the first evaluation value α, the accuracy of determining whether the object is a living body or not can be improved.

In addition to the first evaluation value α, a second evaluation value β is calculated as an evaluation value by the evaluation value calculation unit. It is an evaluation value obtained by introducing a weighting function w _f (f) for each frequency, and by appropriately setting w _f (f) according to an assumed camouflage or the like, the accuracy of determining whether or not the object is a living body can be improved.

1 is a schematic block diagram showing an example of a living body detection device according to an embodiment of the present invention; FIG. 1 is a schematic diagram showing an example of a CSRR-BPF according to an embodiment of the present invention, where (a) shows the top side and (b) shows the bottom side; FIG. FIG. 2 is an enlarged schematic diagram showing a part of the CSRR-BPF according to the embodiment of the present invention, where (a) shows part A surrounded by a two-dot chain line in FIG. A part B surrounded by a dotted chain line is shown. 5 is a graph showing pass characteristics |S21| of a CSRR-BPF according to an embodiment of the present invention and a conventional CSRR-BPF in which two sets of CSRRs are arranged in series; 5 is a graph showing the results of an experiment conducted to compare a CSRR-BPF according to an embodiment of the present invention and a conventional CSRR-BPF in which two sets of CSRRs are arranged in series; (a) is a graph regarding CSRR-BPF according to an embodiment of the present invention, and (b) is a graph regarding CSRR-BPF in which two sets of conventional CSRRs are arranged in series. FIG. 10 is a graph showing experimental results for comparing a first evaluation value α according to the embodiment of the present invention and similarity, which is a conventional evaluation value; FIG. (a) is a graph related to the first evaluation value α according to the embodiment of the present invention, and (b) is a graph related to the similarity which is the conventional evaluation value. 7 is a graph showing results of an experiment conducted to compare a second evaluation value β according to the embodiment of the present invention and an average difference, which is a conventional evaluation value; (a) is a graph regarding the second evaluation value β according to the embodiment of the present invention, and (b) is a graph regarding the average difference which is the conventional evaluation value.

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of a living body detection device according to the present invention will be described below with reference to the drawings. As in Patent Document 1 described above, the living body detection device 1 according to the present invention determines whether or not the detection target part of the human body to be tested is a living body in order to prevent "spoofing" by a forgery. It is mainly used together with biometric authentication devices such as fingerprint authentication and vein authentication. As shown in FIG. 1, the living body detection device 1 according to the present invention has a BPF (Band Pass Filter) in which a CSRR (Complementary Split Ring Resonator) is arranged. hereinafter referred to as CSRR-BPF) 2, a measurement unit 3 that measures the electromagnetic wave response characteristics when the detection target part is brought close to or in contact with the CSRR-BPF 2, and the electromagnetic wave response characteristics measured by the measurement unit 3 are used. and a computer 4 that performs arithmetic processing and the like for biometric determination. The CSRR-BPF2 according to the present invention is characterized by having a structure in which two sets of CSRRs are coaxially arranged as shown in FIGS. The CSRR is composed of two large and small split ring resonators (SRR) arranged coaxially so that the inner and outer split parts are reversed, and as shown in FIG. is one set of CSRR23, and the pair of SRR24 and SRR25 is one set of CSRR26. In each CSRR, the width of the outer and inner SRRs and the width between the two SRRs are equal, and the width of the split is also equal. For example, in the case of CSRR23, the dimensions a13, a14 and a15 are equal, and the dimensions a9 and a12 are equal. In this embodiment, as shown in FIG. 1, only one CSRR-BPF 2 is arranged. Alternatively, the electromagnetic wave response characteristics of the detection target sites of a plurality of subjects may be measured simultaneously.

The CSRR-BPF2 is designed to have a predetermined center frequency according to the information desired to be obtained for the detection target part of the subject (human body), and has a CSRR that is strongly affected by the adjacent medium. When a medium is brought close to the CSRR, the distribution of the electromagnetic field near the CSRR-BPF2 changes due to the influence of the medium. Therefore, it is possible to obtain the response of the electromagnetic field in the depth direction of the medium, that is, the electromagnetic wave response characteristic.

A detection target part includes, for example, a human finger for vein authentication or fingerprint authentication. A human finger is considered to be a structure in which muscle, fat, and skin are layered around a central bone. By defining a human finger as an approximate uniform medium using the volume ratio of each layer, and calculating the skin depth of the human finger based on its dielectric constant and conductivity, it is about 10 mm at 6 GHz and about 10 mm at 10 GHz. At 5 mm and 16 GHz, it is about 2.5 mm, and as the frequency increases, the electromagnetic field penetrating the human finger in the depth direction attenuates. Therefore, when designing the CSRR-BPF2 of the living body detection device 1 used together with vein authentication, the vein authentication uses the three-dimensionally distributed blood vessel pattern as authentication information, and the camouflage is a three-dimensional structure that imitates the blood vessel pattern. Therefore, it is considered effective to set the center frequency to the low frequency side where the electromagnetic wave response characteristic reflecting the characteristics of the entire human finger can be obtained. On the other hand, when designing CSRR-BPF2 to be used in combination with fingerprint authentication, the fingerprint authentication uses the pattern on the surface of the tip of the human finger as authentication information, and the camouflage is a thin film that imitates the pattern and is attached to the surface of the human finger. Therefore, it is considered effective to set the center frequency to the high frequency side where electromagnetic wave response characteristics reflecting the surface characteristics of the human finger can be obtained. Further, the detection target part is not limited to the finger, and other parts may be used. In this case, the CSRR-BPF2 may be designed with a center frequency at which appropriate characteristics can be obtained for that part.

In the present embodiment, as in Patent Document 1, the CSRR-BPF 2 is designed with a central frequency of 10 GHz, assuming the biometric detection device 1 used in combination with fingerprint authentication. Its structure is shown in FIG. FIG. 2(a) is a diagram viewed from the top side, and FIG. 2(b) is a diagram viewed from the bottom side. FIG. 3 is an enlarged view of a part of CSRR-BPF2 in FIG. 2, and FIG. FIG.3(b) is the figure which showed the B section enclosed with the two-dot chain line in FIG.2(b). As described above, the skin depth of a human finger at 10 GHz is about 5 mm, which is about half assuming that the thickness of a human finger is about 10 mm. It is considered that the electromagnetic wave response characteristic reflecting the influence of the structure in the depth direction from the surface to about half of the human finger can be obtained. Due to the layered structure of the human finger, even the electromagnetic wave response characteristics up to half the depth of the human finger include the surface skin, fat, muscle, and even the bone at the core. It can be expected that electromagnetic wave response characteristics that reflect the influence of the structure will be obtained. Therefore, the CSRR-BPF2, which is designed to have a center frequency of 10 GHz, can acquire electromagnetic wave response characteristics that reflect the effects unique to human fingers while being relatively strongly affected by the surface portion, and is therefore suitable for use with fingerprint authentication.

This CSRR-BPF2, for example, similarly to the CSRR-BPF of Patent Document 1, has a glass hardening PPO resin R-4726 (relative permittivity ε _γ =3.4, dielectric loss tangent tan δ=0) which is a dielectric substrate as a substrate. .005, substrate thickness: 1.0 mm) can be used. Copper foils are attached to both surfaces of the dielectric substrate, which can be produced by cutting the transmission line and CSRR structure as shown in FIG. The hatched portions in FIGS. 2 and 3 are copper foils. An example of the dimensions of each part of CSRR-BPF2 designed with a center frequency of 10 GHz in this embodiment is a1=26.0 mm, a2=19.5 mm, a3=10.6 mm, a4 = 4.8 mm, a5 = 10.6 mm, a6 = 7.35 mm, a7 = 4.8 mm, a8 = 7.35 mm, a9 = 0.4 mm, a10 = 0.2 mm, a11 = 0.2 mm, a12 = 0 .4mm, a13 = 0.4mm, a14 = 0.4mm, a15 = 0.4mm, a16 = 0.3mm, a17 = 0.2mm, a18 = 0.2mm, a19 = 0.2mm, b1 = 11.6mm , b2 = 11.6 mm, b3 = 8.55 mm, b4 = 2.4 mm, b5 = 8.55 mm, b6 = 0.4 mm, b7 = 1.2 mm, b8 = 1.2 mm, b9 = 1.6 mm, b10 = 1.6 mm. When measuring the electromagnetic wave response characteristics, it is necessary to cover the CSRR part of the CSRR-BPF 2 with the detection target part, but the CSRR-BPF having a structure in which two sets of CSRRs described in conventional Patent Document 1 are arranged in series In the case of , the length from end to end of the two sets of CSRRs is 6.3 mm in a design with a center frequency of 10 GHz as in this embodiment, whereas in this embodiment, one Two sets of CSRRs fit in a square of 4.8 mm square, and the CSRR part, that is, the mounting area is reduced by 24% compared to the conventional case. Reducing the mounting area makes it possible to locally acquire the difference in the effect when the medium is brought close to CSRR-BPF2, leading to a reduction in false detections caused by insufficient adhesion of the medium and an improvement in detection accuracy. .

Further, ports P1 and P2 provided at both ends of the CSRR-BPF 2 are provided with connectors (not shown), respectively, which are connected to the measurement unit 3 via coaxial cables 5 as shown in FIG. there is In addition, a polypropylene film having a thickness of about 0.2 mm may be placed on the CSRR-BPF 2 in order to flatten the steps of the CSRR due to abrasion or the like and to prevent corrosion of the copper foil.

The measurement unit 3 is composed of, for example, a vector network analyzer or the like, generates an electromagnetic wave for measurement, enters the CSRR-BPF 2, and brings the detection target site close to or in contact with the CSRR-BPF 2. Measure the electromagnetic wave response characteristics at the time. As the electromagnetic wave response characteristics, when the input/output terminals of the CSRR-BPF2 are two ports P1 and P2 as in this embodiment, the transmission characteristic S21 for incident light to the port P1 and passing to the port P2, and the It is possible to obtain four characteristics: a transmission characteristic S12 for passing through the port P1, a reflection characteristic S11 for incident on the port P1 and reflected to the port P1, and a reflection characteristic S22 for incident on the port P2 and reflected to the port P2. I can.

A measured value of the electromagnetic wave response characteristic measured by the measuring unit 3 is input to the computer 4 . The computer 4 uses the input measured value of the electromagnetic wave response characteristic to perform arithmetic processing and the like for determining whether the detection target site is a living body. For example, a CPU (Central Processing Unit: It includes a central processing unit 6, a storage unit 7, an arithmetic processing unit 8, a display unit 10, an operation unit 11, and the like. These units are connected to a system bus 12 as shown in FIG.

The storage unit 7 can store the measured values of the electromagnetic wave response characteristics input from the measurement unit 3, and stores a processing program or the like for determining whether the detection target site is a living body or not. Stored. Although the storage unit 7 is provided in this embodiment, another external computer-readable storage medium may be used.

Based on the processing program stored in the storage unit 7, the arithmetic processing unit 8 performs arithmetic processing and the like for determining whether or not the detection target site is a living body under the control of the CPU 6. FIG. The evaluation value calculation unit 81 of the calculation processing unit 8 calculates an evaluation value using the measured value of the electromagnetic wave response characteristic of the detection target site and the reference value of the living body, and based on the calculated evaluation value, the calculation processing unit 8 determines whether or not the part to be detected is a living body.

The display unit 10 is composed of, for example, a liquid crystal display or the like, and displays the result of living body determination or the like. Instead of the display unit 10, or together with the display unit 10, the result of the biometric determination may be notified by voice. The operation unit 11 is composed of a mouse, a keyboard, and the like, and is used by an operator to input various data, operation commands, and the like. Note that the display unit 10 and the operation unit 11 may be integrated as a touch panel.

The flow of living body detection processing by this living body detection device 1 will be described below. First, the subject to be detected brings the detection target region close to or in contact with the CSRR-BPF2. At that time, the CSRR part of CSRR-BPF2 is covered with the detection target part. Then, the measurement unit 3 measures the electromagnetic wave response characteristic S(f) in a predetermined frequency band. In this embodiment, CSRR-BPF2 designed with a center frequency of 10 GHz for fingerprint authentication is used, the detection target part is the part from the first joint of the right index finger, and the electromagnetic wave response characteristic is the pass characteristic S21 in 3 GHz to 16 GHz. The absolute value |S21| of is used. However, not limited to this, any finger may be used as a detection target portion, and a portion other than a finger may be used if not for fingerprint authentication. In that case, CSRR-BPF2 should be designed with a center frequency according to the information to be obtained. Further, the electromagnetic wave response characteristic is not limited to the transmission characteristic |S21|, and other transmission characteristic |S12|, reflection characteristic |S11|, or reflection characteristic |S22| may be used. Also, a plurality of these may be used. Also, the frequency band for measuring the electromagnetic wave response characteristics may be changed according to the characteristics of the designed CSRR-BPF2.

A measured value S(f) of the electromagnetic wave response characteristic measured by the measuring unit 3 is input to the computer 4 . The evaluation value calculation unit 81 of the arithmetic processing unit 8 of the computer 4 calculates an evaluation value using the input measurement value S(f) and the reference value T(f) of the electromagnetic wave response characteristics of the living body.

The reference value T(f) of the electromagnetic wave response characteristic of the living body is calculated by the evaluation value calculation unit 81, and in the present embodiment, is measured a plurality of times in advance under the same measurement conditions as the measurement value S(f), and stored in the storage unit 7. Using the measured values of the electromagnetic wave response characteristics of the detection target portion of the subject stored in , the standard value T(f) is calculated by averaging them for each frequency. At this time, the standard deviation σ(f) of the reference value T(f) is also calculated at the same time. A fixed number of measurements of the electromagnetic wave response characteristics may be used to calculate the reference value T(f). Since it is considered to lead to a reduction in detection, every time a living body determination is performed, if it is determined to be a living body after the living body determination, the measured value S(f) is stored in the storage unit 7, and the reference value from the next time onwards. It may be added to the measured value used to calculate T(f). At that time, for example, when fingerprint authentication is used together, it is considered that the living body detecting device 1 determines whether or not the body is a living body after it is determined that the fingerprints match in the fingerprint authentication. The measured values of the response characteristics may be linked and stored in the storage unit 7, for example. Also, here, the reference value T(f) is obtained by averaging the measured values of the electromagnetic wave response characteristics, but it is not limited to this, and other statistical values such as the median value may be used. In addition, here, the reference value T(f) is calculated using the measurement value of the detection target region of the subject that has been measured in advance. If so, measured values of electromagnetic wave response characteristics of other fingers such as the middle finger and the ring finger may be used. In this case, without storing the measured value for calculating the reference value T(f) in the storage unit 7, a plurality of CSRR-BPFs 2 are arranged to measure the electromagnetic wave response characteristic S(f) of the detection target site. At the same time, the reference value T(f) may be calculated by measuring the electromagnetic wave response characteristics of the non-detection target portion used for the reference value T(f). In addition, the electromagnetic wave response characteristics of the detection target parts of a plurality of persons different from the subject are measured in advance, the measured values are stored in the storage unit 7, and the reference value T(f) is calculated using these measured values. You may However, in order to improve the discrimination accuracy, it is preferable to calculate the reference value T(f) using the measured values of the electromagnetic wave response characteristics of the same detection target portion of the same subject as in the present embodiment.

The evaluation value calculator 81 calculates a first evaluation value α represented by the following equation (1) as an evaluation value. Here, k(f) is the consistency obtained by dividing the difference between the reference value T(f) and the measured value S(f) expressed by Equation (2) by the standard deviation σ(f) of the reference value T(f) is an index, w _v (k(f)) is a weighting function that performs weighting according to the consistency index k(f) represented by Equation (3), p is a predetermined threshold, and w _f (f) is a weighting function that weights each frequency, f _L is the lower limit in the frequency band to be used, and f _H is the upper limit. First, the matching weight w _v (k(f)) is calculated by equations (2) and (3). The consistency weighting function w _v (k(f)) takes a value from 0 to 1, and the closer to 1, the closer the measured value S(f) is to the reference value T(f), If the difference between the measured value S(f) and the reference value T(f) is within the range of p times the standard deviation σ(f) of the reference value T(f), it is determined that there is agreement, and the agreement weight It is assumed that w _v (k(f))=1, and if it is exceeded, w _v (k(f))=0 as there is no match. Even if it is the same detection target part of the same subject, the measured values of the electromagnetic wave response characteristics do not match completely, and the condition of the subject such as the sweating state at the time of measurement and the proximity of the detection target part to CSRR-BPF2 It has a certain variance depending on how it is made. Therefore, even if the measured value S(f) is obtained from the same detection target site of the same subject as the reference value T(f), the value may deviate from the reference value T(f). , the threshold value p determines how far apart the value is considered to be the same as the reference value T(f). In this embodiment, p=2.7 is used. This value is such that when the measured values of the electromagnetic wave response characteristics of the detection target part of the subject used for calculation of the reference value T(f) follow a normal distribution, the measured value S(f) is the reference value T(f ), the measured value S(f) falls within the range of w _v (k(f))=1 with a probability of 99.3%. If the value of the threshold p is too large, the probability that the measured value S(f) of the counterfeit object is considered to match the reference value T(f) increases. A value between p=2.0 and 3.0 is preferable because the probability that the measured value S(f) of the living body is considered to be inconsistent with the reference value T(f) increases. In addition, in the present embodiment, equation (3) is used to calculate the matching weight function w _v (k(f)). By doing so, it is possible to improve the discrimination accuracy. Then, according to the equation (1), the weight w _v (k(f)) of consistency is multiplied by the weight w _f (f) for each frequency, and the arithmetic mean value in the frequency band used is set as the first evaluation value α calculate. Here, the weighting function w _f (f) that performs weighting for each frequency reduces the influence on the evaluation value in the frequency band where the difference in electromagnetic wave response characteristics between the living body and the camouflage is not seen so much that the difference due to the camouflage is reduced. It is used for emphasizing , and it is possible to improve the discrimination accuracy by setting appropriately according to the type and thickness of the assumed camouflage. w _f (f) takes a value from 0 to 1. For example, w _f (f)=1 in the frequency band where the difference is desired to be emphasized, and w _f (f)=1 in the frequency band where the difference is small and the effect on the evaluation value is to be eliminated. =0. Since both the matching weight function w _v (k(f)) and the weight function w _f (f) for each frequency are functions taking values from 0 to 1, the first evaluation value α also ranges from 0 to 1. The closer the value is to 1, the higher the possibility that the measured value S(f) is caused by the living body.

On the other hand, conventionally, as described in Patent Literature 1, similarity is calculated and used as an evaluation value. Similarity is calculated by the following formula (5). Here, S(f) is the measured value of the electromagnetic wave response characteristic measured by the measuring unit 3, T(f) is the reference value, _fH is the upper limit of the frequency band to be used, and _fL is the lower limit. It is the same as that used for the evaluation value α. S _norm (f) is represented by Equation (6), and is a value obtained by normalizing the measured value S (f) so that the minimum value of S _norm (f) and the minimum value of the reference value T (f) match. D _n is expressed by Equation (7) and is the minimum difference between the measured value S(f) and the reference value T(f) in the frequency band used. As can be seen from these formulas, the degree of similarity does not take into account variations in the measured values of the electromagnetic wave response characteristics used to calculate the reference value T(f). It was an evaluation value that was difficult to appear. On the other hand, the first evaluation value α calculated in the present invention allows for variations in the measured values of the electromagnetic wave response characteristics in the case of a living body without a camouflage attached, as described above. Since the evaluation value emphasizes and extracts the change in the measured value of the electromagnetic wave response characteristics in a certain case, the discrimination accuracy is improved even for extremely thin camouflage objects.

In addition to the first evaluation value α, the evaluation value calculation unit 81 also calculates a second evaluation value β represented by the following equation (4) as an evaluation value. The second evaluation value β is a weighting function w _f (f) is introduced. The average difference is an evaluation value that averagely determines the difference in the graph curve between the measured value S (f) and the reference value T (f). , and the larger the value, the higher the possibility that the measured value S(f) is due to a counterfeit.

Based on two evaluation values, the first evaluation value α and the second evaluation value β calculated by the evaluation value calculation unit 81, the living body determination unit 82 determines whether or not the detection target site is a living body. make a judgment. In the living body determination unit 82, for example, threshold values are set for the first evaluation value α and the second evaluation value β, respectively, and the first evaluation value α and the second evaluation value β are within each threshold value. By determining whether or not the evaluation value β of 2 is included, it is determined whether or not the object is a living body. That is, when both the first evaluation value α and the second evaluation value β calculated by the evaluation value calculation unit 81 are within the threshold value, it is determined that the detection target site is a living body. In this case, it is determined that the detection target site is not a living body. It should be noted that the living body determination may be performed using only one of the first evaluation value α and the second evaluation value β. It is preferable to use two evaluation values, such as an evaluation value β of two.

The flow of the living body detection processing by the living body detection device 1 according to the present embodiment has been described above. An experiment was conducted to evaluate how much the discrimination accuracy is improved compared to the conventional method by using the CSRR-BPF 2 and the first evaluation value α and the second evaluation value β calculated by the evaluation value calculation unit 81. Show the results.

First, experimental results for CSRR-BPF2 having a structure in which two pairs of CSRRs according to the present invention are coaxially arranged are shown. The structure of CSRR-BPF2 of the present invention is shown in FIGS. 2 and 3. FIG. In the conventional CSRR-BPF, which has a structure in which two sets of CSRRs are arranged in series, when measuring the electromagnetic wave response characteristics of the detection target part, both of the two sets of CSRRs arranged in series are covered with the detection target part. However, in the CSRR-BPF2 according to the present invention, two sets of CSRRs are arranged in one place, so that it is easy to cover them with the part to be detected. Further, as described above, the size of the CSRR portion of the CSRR-BPF2 (coaxial type 10 GHz CSRR-BPF2) having a structure in which two sets of CSRRs designed with a center frequency of 10 GHz according to the present invention are arranged on the same axis is 4.8 mm square. On the other hand, in the conventional CSRR-BPF (series type 10 GHz CSRR-BPF) having a structure in which two sets of CSRRs designed with a center frequency of 10 GHz are arranged in series, the length from end to end of the two sets of CSRR parts is The dimension is 6.3 mm, and the placement area of the part to be detected is reduced by about 24%.

FIG. 4 shows measurement results of pass characteristics |S21| of the coaxial 10 GHz CSRR-BPF2 of the present invention and the conventional series 10 GHz CSRR-BPF. The solid line is the measurement result of the coaxial 10 GHz CSRR-BPF2 of the present invention, and the bandwidth is approximately 12 GHz from 4 GHz to 16 GHz. The dotted line is the measurement result of the conventional series type 10 GHz CSRR-BPF, and the bandwidth is approximately 4 GHz from 8 GHz to 12 GHz. It can be seen that the operating frequency band is widened.

Furthermore, FIG. 5 shows the results of living body detection experiments using the coaxial 10 GHz CSRR-BPF2 of the present invention and the conventional series 10 GHz CSRR-BPF. For 20 subjects, measurements were made 10 times for each person using the right index finger without a camouflage attached as a human finger, and the electrical constant of the skin on the surface of the human finger, which is a camouflage, was simulated as a camouflage finger. A skin phantom having a thickness of 0.3 mm was processed into a sheet, which was attached to the index finger of the right hand and measured five times for each person. Here, the conventional similarity and average difference are used for the evaluation value calculated by the evaluation value calculation unit 81 . This is to compare and evaluate how much the discrimination accuracy is improved by the CSRR-BPF2 according to the present invention and the conventional one. The reference value T(f) is obtained by averaging five measurements of the human finger of all subjects, and the measurement value S(f) of the detection target is used to calculate the reference value T(f). Five human finger measurements and sham finger measurements were used. Also, from FIG. 4, the coaxial 10 GHz CSRR-BPF2 of the present invention had the minimum pass characteristic |S21| at 16 GHz, so the frequency band for searching the minimum value used for similarity calculation was calculated with 5 to 16 GHz. Both the present invention and the conventional method were measured by placing a polypropylene film with a thickness of 0.2 mm on the surface of the CSRR-BPF. FIG. 5(a) shows the results of the coaxial 10 GHz CSRR-BPF2 of the present invention, where the vertical axis represents the degree of similarity and the horizontal axis represents the average difference. . FIG. 5(b) shows the result of the conventional serial type 10 GHz CSRR-BPF, where the vertical axis represents the similarity and the horizontal axis represents the average difference. Looking at FIG. 5, it can be seen that the coaxial type 10 GHz CSRR-BPF2 of the present invention separates the plots of the human finger and the fake finger more than the conventional serial type 10 GHz CSRR-BPF. In order to do so, the false acceptance rate (FRR) when the false acceptance rate (FAR) for judging a fake as a living body is set to 0%, and the false rejection rate (FRR) for when the FRR is set to 0%. Evaluation was made using FAR and a value (Equal Error Rate, EER) at which FAR and FRR are equal. The results are shown in Table 1 below. Here, the FRR when the FAR is set to 0% indicates the rate at which the human finger is not accepted when a threshold is set to reject all the fake fingers used in the measurement. Specifically, the minimum value of the average difference and the maximum value of similarity of the fake finger are used as thresholds, and the ratio of human fingers that do not fall within the region below the minimum value of the average value of fake fingers and above the maximum value of similarity is. Further, the FAR when the FRR is 0% indicates the rate at which the fake finger is accepted when a threshold value is set to accept all the human fingers used in the measurement. Specifically, using the maximum value of the average difference and the minimum value of the similarity of human fingers as thresholds, the camouflaged object that falls within the region below the maximum value of the average difference between human fingers and above the minimum value of similarity percentage. The lower these indexes, the higher the discrimination accuracy. From Table 1, the coaxial 10 GHz CSRR-BPF2 according to the present invention is more significant than the conventional series 10 GHz CSRR-BPF in terms of FRR (FAR = 0%), FAR (FRR = 0%), and EER. It can be seen that the discrimination accuracy is improved.

Next, FIG. 6 shows the results of an experiment for evaluating the first evaluation value α calculated by the evaluation value calculation section 81 according to the present invention. Here, the first evaluation value α of the present invention is compared with the similarity, which is the conventional evaluation value, to evaluate how much the discrimination accuracy is improved by the first evaluation value α. A series type 10 GHz CSRR-BPF similar to the conventional one was used. The first evaluation value α is calculated by equations (1) to (3) as described above, where the threshold value p is p=2.7 and the weight for each frequency w _f (f) is the frequency to be used w _f (f)=1 over the entire band. A skin phantom with a thickness of 0.5 mm processed to dimensions of 22.0 mm×14.0 mm was used as a camouflage to be worn on a human finger. FIG. 6A shows the result of using the first evaluation value α of the present invention, where the vertical axis represents the first evaluation value α and the horizontal axis represents the average difference. FIG. 6B shows the result of using similarity, which is a conventional evaluation value, where the vertical axis represents the similarity and the horizontal axis represents the average difference. In FIG. 6, the open plots indicate human fingers, and the black plots indicate fake fingers. As shown in FIG. 6, the similarity, which is the conventional evaluation value, is close to 1 for the human finger, while the maximum value for the fake finger is 0.988, and most of the values are 0.9 or higher. However, with the first evaluation value α of the present invention, the human finger has a value close to 1, while the fake finger has a maximum value of 0.692, indicating that the human finger and the fake finger are very well matched. can be distinguished, and it can be seen that the discrimination accuracy is improved.

Next, FIG. 7 shows the results of an experiment for evaluating the second evaluation value β calculated by the evaluation value calculation section 81 according to the present invention. In this case, CSRR-BPF2 was also measured using a conventional series type 10 GHz CSRR-BPF. As described above, the second evaluation value β is calculated by Equation (4), and is an evaluation value obtained by introducing the weight w _f (f) for each frequency into the average difference, which is the conventional evaluation value. Silicon rubber processed to a size of 22.0 mm×14.0 mm was used as the camouflage to be worn on the finger of the human body. In particular, when a 0.1 mm thick silicon rubber is used as the camouflage, the difference from the human finger becomes small in the high range of the pass characteristics |S21|. Therefore, w _f (f)=0 at 12-16 GHz and w _f (f)=1 at 5-12 GHz to emphasize the difference due to the impersonator in the frequency band below 12 GHz. FIG. 7A shows the result of using the second evaluation value β of the present invention, with the second evaluation value β on the horizontal axis and the first evaluation value α on the vertical axis. FIG. 7B shows the result of using the average difference, which is the conventional evaluation value, with the average difference on the horizontal axis and the first evaluation value α on the vertical axis. Since it is difficult to see the difference just by looking at FIG. 7, in order to perform a quantitative evaluation, the above-described FAR when the FRR is set to 0% and the FRR when the FAR is set to 0% were used for evaluation. The results are shown in Table 2 below. By introducing an appropriate weight w _f (f) using the second evaluation value β according to the present invention, the FRR (FAR = 0%) is improved from 5.2% to 1.7%, and the discrimination accuracy is I can see that it is improving.

As described above, in the CSRR-BPF 2 of the living body detection device 1 according to the present invention, and in all of the first evaluation value α and the second evaluation value β calculated by the evaluation value calculation unit 81, each of the conventional Patent Document 1 It contributes to the improvement of discrimination accuracy compared to that used in the described living body detection device, and by using the living body detection device 1 according to the present invention, living body detection with high discrimination accuracy can be performed.

The embodiments of the present invention are not limited to those described above, and can be modified as appropriate without departing from the spirit of the present invention.

INDUSTRIAL APPLICABILITY The biometric detection device according to the present invention can be effectively used as highly reliable information security technology by combining it with personal biometric authentication technology such as fingerprint authentication and vein authentication.

1 living body detection device 2 CSRR-BPF
3 measurement unit 7 storage unit 81 evaluation value calculation unit 82 biological determination unit

Claims (3)

a band-pass filter having a split-ring resonator designed to have a predetermined center frequency according to the detection target site of the subject;
a measurement unit that generates an electromagnetic wave and measures an electromagnetic wave response characteristic in a predetermined frequency band when the detection target portion is brought close to or in contact with the band-pass filter;
an evaluation value calculation unit that calculates an evaluation value using the measured value of the electromagnetic wave response characteristic measured by the measurement unit and a reference value of the electromagnetic wave response characteristic of a living body;
a living body determination unit that determines whether the detection target part is a living body based on the evaluation value calculated by the evaluation value calculation unit,
The evaluation value calculation unit
S(f) is the measured value of the electromagnetic wave response characteristic measured by the measuring unit, T(f) is the reference value of the electromagnetic wave response characteristic of the living body, σ(f) is the standard deviation of T(f), and the following Let k(f) be the consistency index represented by Equation (2), w _v (k(f)) be the weighting function corresponding to k(f), w _f (f) be the weighting function for each frequency, and When the lower limit of the frequency band to be used is f _L and the upper limit is f _H ,
A living body detecting device, wherein a first evaluation value α represented by the following equation (1) is calculated as the evaluation value.

The weight function w _v (k(f)) is
When p is a predetermined threshold,
2. The living body detecting device according to claim 1 , wherein the function is represented by the following formula (3).

The evaluation value calculation unit
3. The living body detection according to claim 1 , wherein as the evaluation value, in addition to the first evaluation value α, a second evaluation value β represented by the following equation (4) is calculated. Device.

Images