JP2012133683A - Personal authentication device and personal authentication system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve precision of authentication of an individual with a small quantity of learning and without a special operation.SOLUTION: A personal authentication device (C, NK) includes a pressure detection member (SN) which has a first detection part (SN1) and a second detection part (SN1); feature quantity arithmetic means (C5B) of computing a first feature quantity (u) and a second feature quantity (u); correlation coefficient storage means (C2M1) of associating and storing a registrant and a correlation coefficient (r); distance arithmetic means (C5C) of computing a Mahalanobis distance (D) of a subject person based upon the correlation coefficient (r), first feature quantity (u), and second feature quantity (u); threshold determination means (C5D) of determining the Mahalanobis distance (D) is equal to or less than a threshold (S); registrant authentication means (C5E) of authenticating the subject person as a registrant associated with the correlation coefficient (r) used to compute the Mahalanobis distance (D); and authentication display means (C5H, C5H').

Description

  The present invention relates to a personal authentication device and a personal authentication system.

In recent years, with the spread and development of computer systems and security systems, the use of personal authentication technology for authenticating individuals (users, users) such as users and administrators has been expanded. In particular, biometrics authentication, which authenticates individuals based on physical characteristics such as fingerprints, veins, and faces, has been widely used for entrance / exit management in condominiums, authentication methods for cash cards and passports, etc. is doing. However, the biometric authentication requires a user to perform a predetermined operation such as touching a reading sensor with a finger or holding a hand to read a fingerprint, a vein, or the like. For this reason, research and development of personal authentication technology that can authenticate a user without performing any special work / operation are being actively conducted.
Here, for example, techniques described in Patent Documents 1 and 2 and Non-Patent Document 1 described below are known as techniques related to a personal authentication technique based on a user's natural behavior.

  Japanese Patent Application Laid-Open No. 2007-179422 as Patent Document 1 is provided inside the seat (11) of the chair (1) on which the user sits when using a PC (Personal Computer) (2). A technique is described in which a pressure sensor (15) is connected to the PC (2) and whether or not there is a seated person based on a signal from the pressure sensor (15) is described. Further, in Patent Document 1, when it is determined that the user has left the seat, the use of the PC (2) is restricted, and the pressure created based on the signal when the seated person is seated. It is determined whether or not the distribution data and the pressure distribution data of the user stored in advance match within a predetermined error range, and if the seated person is the user, use of the PC (2) A technique for releasing the restriction is described.

In Patent Document 1, the depth distribution of the seat portion (11) and the width direction perpendicular thereto are subdivided into a plurality of regions, and the pressure distribution is represented by data indicating how much pressure is applied to which region. Data is structured.
That is, in Patent Document 1, the pressure distribution data from the pressure sensor (15) and the personal pressure distribution data stored in advance are collated as sequence data of pressure values in each region, and within a predetermined error range. In the case of a match, for example, when the difference value of the numerical value of each area is within a predetermined error range, or when the number of areas where the numerical values do not match is within a predetermined error range, a technique that can be authenticated as the user Are listed.

Non-Patent Document 1 discloses 33 types of pressure values from 32 sensors attached to the seat surface of a chair and a total value of each pressure value by a support vector machine (SVM). A technique for authenticating an individual from each measurement data is described. In Non-Patent Document 1, when the support vector machine learns the characteristics of each individual by 19 sets of trials, that is, when there are 19 types of learning data (training samples), for example, a general family is assumed. In the sample in which 5 students (male students of the laboratory) are seated and stable, each individual can be identified with an identification rate of 98.3 [%], and 10 students (laboratory) assuming a small office It is described that each individual could be identified with an identification rate of 96.6 [%] even in a sample in which the male student was seated and stable.
Here, the support vector machine is a pattern discriminator using a neural network, and the sum of values of a plurality of linear input elements (neurons) multiplied by a connection load (synaptic load) exceeds a preset threshold value. It is a layered neural network that outputs an output value indicating whether or not.

Specifically, an input feature vector as a set (input value) of n element values of the input layer is x (x = (x1, x2,..., Xn)), and one element of the output layer is The value (output value) is y, the coupling load vector (coupling load vector) between each element of the input layer and the element of the output layer is ω (ω = (ω1, ω2,..., Ωn)), and the coupling load vector the transpose vector of omega and omega ^T, if the threshold has been is h, the output value y is indicated by the following equation (1).
y = sign (ω ^T x−h) (1)
= Sign (x1ω1 + x2ω2 +... + Xnωn−h)

Sign (ω ^T x−h) is a sign function that becomes +1 when (ω ^T x−h)> 0 and becomes −1 when (ω ^T x−h) ≦ 0. That is, it is a binary function that becomes +1 when the inner product of the vectors x and ω exceeds the threshold value h, and becomes −1 when it does not exceed the threshold value h.
Further, the support vector machine separates the learning data, that is, a pair of input feature vector x and output value y, from each of the data points, that is, the distance (margin) from the input feature vector x. The parameter of the linear input element is learned, that is, the combined weight vector ω is updated based on the criterion of maximizing the margin for obtaining (hyperplane).

  In Japanese Patent Application Laid-Open No. 2007-219974 as Patent Document 2, an individual shoe sole pressure pattern such as a visitor (61) or a staff member (71) is obtained as personal identification information of the individual embedded in a floor surface or a rug. A technique for identifying the individual by a floor sensor (40) having a plurality of sole pressure sensors (41-43) is described. In Patent Literature 2, the personal identification information is acquired by the shoe sole pressure sensor 1 (41) embedded in the floor (63) in front of the reception window (210) and is received and registered in the personal position recognition device (1). Along with the personal identification information acquired by the shoe sole pressure sensors 2 to n (41) embedded in the floor facing the entrance doors (86 to 91) of the rooms (81 to 84) in the management area of the building. Thus, there is described a technique for tracking and recording the behavior of each individual by identifying the movement position of each registered individual.

Further, in Patent Document 2, the shoe sole pressure pattern is generated based on the outer contour line of the shoe sole measured by the shoe sole pressure sensor (41 to 43) in each room and the grounding point not less than a predetermined pressure. A technique for setting identification accuracy by adjusting the predetermined pressure to increase or decrease the number of grounding points is described.
That is, in Patent Literature 2, the shoe sole pressure pattern acquired by the shoe sole pressure sensor (41 to 43) in each room and the individual shoe sole pressure pattern stored in advance at the reception counter (210) are used as image data. A technique is described in which individual verification is performed by searching for an individual that matches the outer contour line and the contact point.

JP 2007-179422 A (Abstract, “0017” to “0023”, FIG. 1, FIG. 3, FIG. 4) JP 2007-219974 A (Abstract, “0024” to “0050”, “0054”, FIGS. 1 to 3)

M. Yamada, 4 others, “Soft authentication and behavior analysis using a chair with sensors attached: hipprint authentication (Soft personal authentication and behavior analysis using a chair with a sensor for seat pressure authentication) ) ”,“ Online ”, 2008, Springer-Verlag,“ November 15, 2010 search ”, Internet <URL: https://www.pervasive.jku.at/Teaching/_2009SS/ SeminarausPervasiveComputing / Begleitmaterial / Related% 20Work% 20% 28Readings% 29 / 2008_Soft% 20authentication% 20and% 20behavior% 20analysis% 20using% 20a% 20chair% 20with% 20sensors% 20attached% 20-% 20hipprint% 20authentication% 20_Kudo.pdf> “NITTA CORPORATION [Product Information] -Tactile Sensor System”, “online”, NITTA CORPORATION, “November 15, 2010 Search”, Internet <URL: http://www.nitta.co.jp/product/ mechasen / sensor / tactile_system_sensor.html> Kouta, 3 others, “Judgment technology for automating sensory tests”, “online”, 2010, Fujitsu Ten Limited, “November 15, 2010 search”, Internet <URL: http: // www .fujitsu-ten.co.jp / gihou / jp_pdf / 55 / 55-4.pdf> Tetsuro Sugihara, “Data Science for Environment and Quality”, “online”, “November 15, 2010 search”, Internet <URL: http://heartland.geocities.jp/ecodata222/index.html>

(Problems of conventional technology)
In Patent Documents 1 and 2, it is determined whether or not the measurement information (measured pressure distribution data) and the stored information (stored pressure distribution data) match (completely match, partially match). For this reason, for example, even the person himself / herself may slightly change the measured pressure distribution data when the position and posture of the pressure sensor at the time of measurement are slightly different from those at the time of registration (during storage). There was a problem that there was a high possibility of misidentifying the person himself / herself.
On the other hand, in Non-Patent Document 1, personal authentication is performed by a support vector machine with a high identification rate of 96 [%] or more from 33 pieces of measurement information.
However, similar to Patent Documents 1 and 2, pressure distribution data, that is, individual authentication is performed by directly inputting each pressure value and total value measured by 32 sensors into a support vector machine, If the measured pressure distribution data changes significantly compared to the time of learning, the learning amount (training sample) in a sitting and stable state may not be sufficient, and the person himself / herself must be the person himself / herself. There was a possibility of misunderstanding.

In Non-Patent Document 1, in order to further increase the identification rate, it is necessary to further increase the number of learning data. However, in a neural network such as a support vector machine, the same calculation is repeated multiple times to learn. However, there is a problem that the amount of calculation for learning tends to increase and the amount of calculation increases as the number of learning data required increases (for example, Japanese Patent Application Laid-Open Nos. 2004-187562, “0026”, “0027”). Etc.)).
Therefore, for example, in order to ensure a high identification rate when the number of sensors increases and the number n of elements in the input layer increases, or when the number of persons subject to personal authentication increases and the number of elements in the output layer increases. If the number of necessary learning data increases, the amount of computation for learning may become enormous.

  An object of the present invention is to improve the accuracy of authentication of individuals who do not perform special operations with a small learning amount.

In order to solve the technical problem, the personal authentication device of the invention according to claim 1 comprises:
A chair on which the subject sits;
A first detector configured to detect a first pressure value, which is a pressure in the first detection region by the subject placed in a first detection region set in advance on the seat surface of the chair and seated; A second detector configured to detect a second pressure value, which is a pressure in the second detection region by the subject who is seated in the second detection region set in advance in a region shifted from the detection region. A pressure sensing member;
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
It is provided with.

The invention according to claim 2 is the personal authentication device according to claim 1,
When registering the registrant, the first pressure value and the second pressure value of the registrant detected by the pressure detection member, the average being 0 and the standard deviation being 1 The correlation coefficient calculated based on a reference space as an authentication standard of the registrant based on a set of combinations of a feature quantity and the second feature quantity having an average of 0 and a standard deviation of 1; The correlation coefficient storage means for storing the registrant in association with each other,
It is provided with.

The invention according to claim 3 is the personal authentication device according to claim 2,
When the registration of the registrant is updated, a combination of the first feature quantity and the second feature quantity calculated by the feature quantity calculation means when the registrant is authenticated by the registrant authentication means is included. Correlation coefficient updating means for associating the correlation coefficient recalculated based on the reference space with the registrant and storing the correlation coefficient in the correlation coefficient storage means;
It is provided with.

In order to solve the technical problem, the personal authentication device of the invention according to claim 4 provides:
The floor on which the subject stands upright,
A first detector configured to detect a first pressure value, which is a pressure of a sole in the first detection area, arranged by the subject standing in an upright position in the first detection area set on the floor; A second detection unit that is arranged in a second detection region set in advance in a region deviated from one detection region and detects a second pressure value that is a pressure of a sole in the second detection region by the subject standing upright. And a pressure detection member having,
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
It is provided with.

In order to solve the technical problem, the personal authentication system according to claim 5 provides:
A chair on which the subject sits;
A first detector configured to detect a first pressure value, which is a pressure in the first detection region by the subject placed in a first detection region set in advance on the seat surface of the chair and seated; A second detector configured to detect a second pressure value, which is a pressure in the second detection region by the subject who is seated in the second detection region set in advance in a region shifted from the detection region. A pressure sensing member;
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
It is provided with.

In order to solve the technical problem, the personal authentication system of the invention according to claim 6 provides:
The floor on which the subject stands upright,
A first detector configured to detect a first pressure value, which is a pressure of a sole in the first detection area, arranged by the subject standing in an upright position in the first detection area set on the floor; A second detection unit that is arranged in a second detection region set in advance in a region deviated from one detection region and detects a second pressure value that is a pressure of a sole in the second detection region by the subject standing upright. And a pressure detection member having,
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
It is provided with.

According to the first and fourth aspects of the present invention, compared to a configuration in which it is not determined whether or not the Mahalanobis distance of the subject is equal to or less than a threshold value, authentication of an individual who does not perform a special operation with a small amount of learning. Accuracy can be improved.
According to the second aspect of the present invention, the correlation coefficient can be calculated based on the first pressure value and the second pressure value for the registrant detected by the pressure detection member.
According to the third aspect of the present invention, the correlation coefficient can be updated based on the latest first pressure value and second pressure value, and the person is not the person due to a change in the physical characteristics of the person over time. Can be reduced.

FIG. 1 is an overall explanatory diagram of a personal authentication system according to a first embodiment, FIG. 1A is a right side view of the automobile of the first embodiment, and FIG. 1B is an enlarged perspective view of a driver's seat of the automobile of the first embodiment. FIG. 2 is an explanatory diagram of the sensor sheet of Example 1, FIG. 2A is an enlarged perspective view of the sensor sheet of Example 1, and FIG. 2B is an enlarged explanatory diagram of the button sensor of Example 1. FIG. 3 is an explanatory diagram illustrating the function of the control unit according to the first embodiment in a block diagram (functional block diagram). FIG. 4 is an explanatory diagram of an example of the pressure distribution on the seating surface output by the sensor seat of the first embodiment. FIG. 5 is an explanatory diagram of a main image according to the first embodiment. FIG. 6 is an explanatory diagram of a registration start image according to the first embodiment. FIG. 7 is an explanatory diagram of an image displayed on the registration start image guidance display unit of the first embodiment, FIG. 7A is an explanatory diagram of a detection start guidance image of the first embodiment, and FIG. 7B is a detection completion guidance of the first embodiment. It is explanatory drawing of an image. FIG. 8 is an explanatory diagram of a registration start image according to the first embodiment. FIG. 9 is an explanatory diagram of an image displaying the authentication result of the first embodiment, FIG. 9A is an explanatory diagram of a success image of the first embodiment, and FIG. 9B is an explanatory diagram of a failure image of the first embodiment. FIG. 10 is a flowchart of the registration process according to the first embodiment. FIG. 11 is a flowchart of the authentication process according to the first embodiment. FIG. 12 is a flowchart of the registration update process of the first embodiment, and is an explanatory diagram of the subroutine of ST106 of FIG. FIG. 13 is a diagram for explaining the operation of the first embodiment. FIG. 13A is a semilogarithmic graph of the authentication result in which the vertical axis is a common logarithm of Mahalanobis distance and the horizontal axis is an equally divided frame for each of five subjects. FIG. 13B is a graph of the authentication result of each subject when subject B is the user when subject A is the user. FIG. 14 is an overall explanatory diagram of the personal authentication system of the second embodiment, and is an explanatory diagram corresponding to FIG. 1B of the first embodiment. FIG. 15 is an explanatory diagram showing the function of the room entry management device of the second embodiment in a block diagram (functional block diagram), and is an explanatory diagram corresponding to FIG. 3 of the first embodiment. FIG. 16 is a flowchart of the authentication process performed by the room entrance management apparatus according to the second embodiment, and is an explanatory diagram corresponding to FIG. 11 according to the first embodiment.

Next, specific examples of embodiments of the present invention (hereinafter referred to as examples) will be described with reference to the drawings. However, the present invention is not limited to the following examples.
In order to facilitate understanding of the following description, in the drawings, the front-rear direction is the X-axis direction, the left-right direction is the Y-axis direction, the up-down direction is the Z-axis direction, and arrows X, -X, Y, -Y, The direction indicated by Z and -Z or the indicated side is defined as the front side, the rear side, the right side, the left side, the upper side, the lower side, or the front side, the rear side, the right side, the left side, the upper side, and the lower side, respectively.
In the figure, “•” in “○” means an arrow heading from the back of the page to the front, and “×” in “○” is the front of the page. It means an arrow pointing from the back to the back.
In the following description using the drawings, illustrations other than members necessary for the description are omitted as appropriate for easy understanding.

FIG. 1 is an explanatory diagram of a personal authentication system according to the first embodiment, FIG. 1A is a right side view of the automobile according to the first embodiment, and FIG. 1B is a perspective enlarged explanatory view of a driver seat of the automobile according to the first embodiment.
FIG. 2 is an explanatory diagram of the sensor sheet of Example 1, FIG. 2A is an enlarged perspective view of the sensor sheet of Example 1, and FIG. 2B is an enlarged explanatory diagram of the button sensor of Example 1.
1, a personal authentication system S according to the first embodiment of the present invention includes a control unit (personal authentication device) of a vehicle V on which a user (registrant, driver, passenger, owner, car sharing member / member) is boarded. , Controller) C. 1A and 1B, the automobile V of the first embodiment is provided with a driver's seat (chair, seating member) CH on which the user is seated. The driver's seat CH has a seating surface CH1 disposed in the lower part on the front side, and a backrest CH2 extending in the rear oblique information from the rear end of the seating surface CH1.

In FIG. 1B and FIG. 2, on the seat surface CH1, a sheet-like sensor sheet (pressure detection member) that is connected to the control unit C and detects the pressure with which the seated user presses the seat surface CH1. , Pressure distribution detection member) SN is laid.
In FIG. 2A, the sensor sheet SN of Example 1 includes 20 squares (cells) in the front-rear direction as an example in the vertical direction, 18 squares in the left-right direction as an example in the horizontal direction, and 360 squares in total (20 × 18 = 360 [mass]) adjacent detection areas (first detection area, second detection area,..., 360 detection area) are set in advance. Further, in FIG. 2B, in the sensor seat SN, there are 360 pieces of pressure (first pressure value, second pressure value,..., 360 pressure value) detected by the seated user in each detection region. A button-shaped button sensor (detection unit, pressure sensor) SN1 is arranged.

Each button sensor SN1 of the first embodiment is configured by so-called pressure-sensitive conductive ink, and a resistance value [Ω] of the contact resistance with the electrode SN2 is set in inverse proportion to the pressure pressed by the user. It has become so. Specifically, the resistance value [Ω] decreases as the pressure value [gf / cm ² ] increases from the no-load resistance value (approximately several M [Ω]) when the user is not seated. It is set to be. That is, the current value [A] flowing (detected or outputted) through the electrode SN2 is set to increase as the pressure value [gf / cm ² ] increases. In Example 1, each button sensor SN1 can detect a pressure value corresponding to the current value [A] in 256 gradations (0 to 255).
The pressure-sensitive conductive ink is known, for example, as described in Non-Patent Document 2 (“NITTA CORPORATION [Product Information] —Tactile Sensor System”).

(Description of the control part C of Example 1)
FIG. 3 is an explanatory diagram illustrating the function of the control unit according to the first embodiment in a block diagram (functional block diagram).
In FIG. 3, the control unit C stores an I / O (input / output interface) that performs input / output of signals to / from the outside, adjustment of input / output signal levels, and the like, a program and data for performing necessary startup processing, and the like. ROM (Read Only Memory), RAM (Random Access Memory) for temporarily storing necessary data and programs, CPU (Central Processing Unit) that performs processing according to the activation program stored in ROM, etc. And a clock oscillator and the like, and various functions can be realized by executing programs stored in the ROM and RAM.
The control unit C stores basic software for controlling basic operations, a so-called operating system OS, a personal authentication program AP1 as an application program, and other software (not shown).

(Signal input element connected to the control unit C of the first embodiment)
FIG. 4 is an explanatory diagram of an example of the pressure distribution on the seating surface output by the sensor seat of the first embodiment.
An output signal such as the next signal output element SN is input to the control unit C.
SN: Sensor sheet The sensor sheet SN outputs the pressure value of the seating surface CH1 by the seated user shown in FIG. 4 by outputting each pressure value for each detection region detected by each button sensor SN1.
(Personal authentication program AP1 of Example 1)
The personal authentication program AP1 has the following functional means (program modules).

FIG. 5 is an explanatory diagram of a main image according to the first embodiment.
C1: Main image display means The main image display means C1 displays the main image 101 relating to the control of the automobile V shown in FIG. 5 on the front monitor FM of the automobile V. In FIG. 5, the main image 101 according to the first embodiment includes a registration start button 101a for starting user registration.
C2: Registration means Registration means (correlation coefficient learning means, learning means) C2 includes registration start determination means C2A, registration start image display means C2B, detection start guide image display means C2C, detection start determination means C2D, Feature quantity calculation means C2E, feature quantity storage means C2F, detection completion guidance image display means C2G, detection completion judgment means C2H, trial number coefficient means C2J, trial completion judgment means C2K, and unit space generation means C2L , A correlation matrix calculation means C2M, a correlation matrix storage means C2N, a registration completion image display means C2P, and a registration confirmation determination means C2Q for registering a user (correlation matrix R in the equation (3) described later) Learn for each user).

FIG. 6 is an explanatory diagram of a registration start image according to the first embodiment.
C2A: Registration Start Determination Unit The registration start determination unit C2A determines whether or not to start user registration by determining whether or not the registration start button 101a is selected.
C2B: Registration Start Image Display Unit The registration start image display unit C2B displays a registration start image 102 for starting registration shown in FIG. 6 on the front monitor FM. In FIG. 6, the registration start image 102 according to the first embodiment includes a guidance display unit 102 a for displaying guidance for registration.

FIG. 7 is an explanatory diagram of an image displayed on the registration start image guidance display unit of the first embodiment, FIG. 7A is an explanatory diagram of a detection start guidance image of the first embodiment, and FIG. 7B is a detection completion guidance of the first embodiment. It is explanatory drawing of an image.
C2C: Detection start guide image display means When the pressure distribution (see FIG. 4) is detected by the sensor sheet SN, the detection start guide image display means C2C guides the user to start the pressure distribution detection shown in FIG. 7A. A detection start guide image 103 is displayed on the guide display unit 102a.
C2D: Detection start determining means The detection start determining means C2D determines whether or not the sensor sheet SN has started detecting pressure distribution. That is, it is determined whether or not one trial (one set of seating / seating of the user) has started on the sensor seat SN. The detection start determination unit C2D according to the first embodiment is configured such that the sensor sheet SN starts detecting the pressure distribution when the pressure values no longer change after the button sensors SN1 of the sensor sheet SN detect the pressure values. It is determined that

C2E: Feature amount calculation means The feature amount calculation means C2E has 35 preset types for specifying the relevance between the pressure values based on the pressure values detected by the button sensors SN1 of the sensor sheet SN. Feature amounts (first feature amount, second feature amount,..., 35th feature amount) that are values of the features (first feature, second feature,..., 35th feature) are calculated.
In the feature amount calculation means C2E of the first embodiment, the following features (1) to (35) are set in advance.

(1) Left side maximum pressure value (maximum value of each pressure value in each detection area of the user's left half)
(2) Right-side maximum pressure value (maximum value of each pressure value in each detection area of the user's right half)
(3) Number of button sensors (detection area, mass, cell) having pressure values of 1 to 63 (4) Number of button sensors having pressure values of 64 to 127 (5) Buttons having pressure values of 128 to 191 Number of sensors (6) Number of button sensors having pressure values of 192 to 255 (7) Number of button sensors having pressure values of 32 to 95 (8) Number of button sensors having pressure values of 96 to 159 (9 ) Number of button sensors having pressure values of 160 to 223 (10) Number of button sensors having pressure values of 1 or more

(11) Total pressure (total value of each pressure value)
(12) Average of each pressure value in the button sensor having one or more pressure values (13) Variance of each pressure value in the button sensor having one or more pressure values (14) One or more pressures in the user's thigh Number of button sensors having value (15) Total pressure in user's thigh (total value of each pressure value)
(16) Average of each pressure value of only the button sensor having one or more pressure values in the user's thigh (17) Variance of each pressure value of only the button sensor having one or more pressure values in the user's thigh

(18) Number of button sensors having one or more pressure values at the user's buttocks (19) Sum of pressures at the user's buttocks (total value of each pressure value)
(20) Average of each pressure value of only the button sensor having one or more pressure values at the user's hip (21) Variance of each pressure value of only the button sensor having one or more pressure values at the user's hip (22 ) Distance from left end to center of gravity of user's left half (distance in squares)
(23) Distance from the rear end to the center of gravity position of the left half of the user (24) Distance from the left end to the center of gravity of the right half of the user (25) Distance from the rear end to the center of gravity of the right half of the user (26) Distance between left and right center of gravity positions (distance from the center of gravity of the user's left half to the center of gravity of the right half)

(27) Distance from the center of gravity of the left half of the user to the left end calculated only by a button sensor (detection area, mass) having a pressure value of 25% or more of the left-side maximum pressure value (28) The left-side maximum pressure value The distance from the center of gravity position of the left half of the user to the rear end calculated by only the button sensor having a pressure value of 25% or more (29) Calculated only by the button sensor having a pressure value of 25% or more of the right-side maximum pressure value Distance from the center of gravity of the user's right half to the left end (30) From the center of gravity of the user's right half calculated only by the button sensor having a pressure value of 25% or more of the maximum right side pressure value to the rear end The distance between the left and right center of gravity calculated by only the button sensor having a pressure value of 25% or more of the left and right maximum pressure values (the user of (27), (28)) From the center of gravity position of the left body (29), the distance to the center of gravity of the right body (30))

(32) Number of button sensors having a pressure value of 1 or more adjacent to a button sensor having a pressure value of 0 in the left half of the user (33) Adjacent to a button sensor having a pressure value of 0 in the right half of the user Number of button sensors having a pressure value greater than or equal to 1 (34) 25% or more of the maximum pressure value on the left side adjacent to a button sensor having a pressure value in the left half of the user that is less than 25% of the maximum pressure value on the left Number of Button Sensors Having a Pressure Value of (35) The pressure value in the right half of the user has a pressure value of 25% or more of the maximum pressure value on the right side adjacent to the button sensor that is less than 25% of the maximum left pressure value Number of button sensors

  Therefore, for example, when the pressure distribution shown in FIG. 4 is output by the sensor sheet SN, (1) the first feature value is “183”, (2) the second feature value is “225”, and (3) the first feature value. 3 feature value is “162”, (4) the fourth feature value is “7”, (5) the fifth feature value is “6”, (6) the sixth feature value is “1”, and (7) the seventh feature. The quantity is “36”, (8) the eighth feature quantity is “9”, (9) the ninth feature quantity is “3”, (10) the tenth feature quantity is “176”, and (11) the eleventh feature quantity is “5239”, (12) the twelfth feature amount is “about 29.767”, (13) the thirteenth feature amount is “about 1243.475”, (14) the fourteenth feature amount is “122”, (15) the number 15 feature quantity is “2178”, (16) 16th feature quantity is “about 17.852”, (17) 17th feature quantity is “about 209.404”, (18) 18th feature quantity. “54”, (19) 19th feature value is “3061”, (20) 20th feature value is “about 56.685”, (21) 21st feature value is “about 250.031”, (22) No. 22 feature quantity is “about 4.939”, (23) 23rd feature quantity is “about 7.335”, (24) 24th feature quantity is “about 12.989”, (25) 25th feature quantity is “ (About 7.545), (26) the 26th feature amount is "about 8.052", (27) the 27th feature amount is "about 5.300", (28) the 28th feature amount is "about 6.411". (29) The 29th feature value is “about 12.807”, (30) the 30th feature value is “about 6.737”, (31) the 31st feature value is “about 7.514”, (32) 32 feature value is “31”, (33) 33rd feature value is “36”, (34) 34th feature value is “16”, and (35) 35th feature value is “15”. It made.

C2F: Feature quantity storage means The feature quantity storage means C2F stores each feature quantity calculated by the feature quantity calculation means C2E. The feature amount storage means C2F according to the first embodiment stores 35 types of feature amounts calculated for each trial as one set. In the feature amount storage means C2F of the first embodiment, the registered user and the calculated feature amount are stored in association with each other. Specifically, when the number of trial completions is MN [times], each feature amount of the total MN set calculated after the trial is completed after the registration start button 101a of the main image 101 is selected is The feature amount of “first user”, the feature amount of “second user”,... Are stored in the order of selection of registration start button 101a (in the order of registration). That is, when the registration number (registration ID), which is the registration order of the user, is p (p = 1, 2,...), “Each feature amount of the MN set” obtained by the user's trial, “Registration number p” is associated and stored.

C2G: Detection completion guide image display means The detection completion guide image display means C2G shows the pressure distribution to the user as shown in FIG. 7B when each feature quantity is stored in the feature quantity storage means C2F and the pressure distribution is detected again. A detection completion guidance image 104 for guiding the completion of detection is displayed on the guidance display unit 102a.
C2H: Detection completion determination unit The detection completion determination unit C2H determines whether or not the detection of the pressure distribution by the sensor sheet SN is completed. That is, it is determined whether or not one trial has been completed and the user has left the sensor sheet SN for the next trial. The detection completion determination unit C2D according to the first embodiment is configured so that the pressure by the sensor sheet SN is detected when each button sensor SN1 of the sensor sheet SN does not detect each pressure value, that is, when all the pressure values become zero. It is determined that the distribution detection is completed.

C2J: Number-of-trials counting unit The number-of-trials counting unit C2J counts the number of trials n (n = 1, 2,..., MN) in which each feature amount is calculated and stored by repeating trials.
C2K: Trial Completion Determination Unit The trial completion determination unit C2K determines whether the number of trials n counted by the trial number coefficient unit C2J has reached a preset maximum value MN [times]. It is determined whether or not the trial for calculating and storing the quantity has been completed.
The registration unit C2 according to the first embodiment determines that the user has left the sensor sheet SN for the next trial by the detection completion determination unit C2H and completes MN trials by the trial completion determination unit C2K. If it is determined that the detection start guide image 103 is displayed by the detection start guide image display means C2C for the next trial, and whether the next trial is started by the detection start determination means C2D Is determined.

C2L: Unit space generation means The unit space generation means (reference space generation means, feature amount standardization means) C2L is determined by the trial completion determination means C2K that the trial for calculating and storing each feature amount has been completed. In this case, a unit space (reference space) is generated as a user authentication standard. The unit space generation unit C2L according to the first exemplary embodiment includes MN sets of feature amounts (MN first feature amounts, MN second feature amounts,..., MN first feature amounts stored in the feature amount storage unit C2F. A unit space is generated based on the combination of (35 feature amounts). Specifically, MN first feature values, MN second feature values,..., MN 35th feature values are normalized (standardized) so that the average is 0 and the standard deviation is 1. , Normalization, and linear transformation with an average of 0 and a standard deviation of 1) to generate a unit space.

  That is, in Example 1, when a combination of the first feature value to the 35th feature value calculated by one trial is a 35-dimensional vector, a 35-dimensional set formed by a set of MN 35-dimensional vectors is configured. A vector space (a partial space whose center is the origin with the average of each feature quantity as a component) is generated as a unit space. In the unit space, the values of each vector component are all zero, that is, the origin is (0, 0,..., 0), and MN values (normalized) when each vector component is a random variable. The distribution of probability (probability distribution) in which the first feature amount, the normalized second feature amount,..., The normalized 35th feature amount) are output is a normal distribution.

C2M: Correlation matrix calculation means Correlation matrix calculation means (correlation coefficient matrix calculation means) C2M is a phase for specifying the correlation between two different types of features based on the unit space generated by the unit space generation means C2L. Correlation coefficient calculation means C2M1 for calculating the number of relationships for each user is provided, and a correlation matrix (correlation coefficient matrix) R, which is a matrix of correlation coefficients, is calculated. In the correlation coefficient calculation means C2M1 of the first embodiment, arbitrary natural numbers from 1 to MN are i and j, and MN i-th feature values (one feature value) are x ₁ , x ₂ ,. _MN , when the MN j-th feature amount (the other feature amount) is y ₁ , y ₂ ,..., Y _MN and the correlation coefficient between the i-th feature amount and the j-th feature amount is r _ij The correlation coefficient r _ij is calculated by the following equation (2).
r _ij = (x ₁ y ₁ + x ₂ y ₂ +... + x _MN y _MN )
/ {(X ₁ ² + x ₂ ² + ... + x _MN ² ) ^1/2
X (y ₁ ² + y ₂ ² +... + Y _MN ² ) ^1/2 } Formula (2)

Therefore, the correlation matrix calculation unit C2M according to the first embodiment uses the correlation coefficient r _{(1) (1)} (r _{(1) ()} between the first feature value and the first feature value calculated by the equation (2). ₁₎ = 1), and the correlation coefficient between the first feature value and the second feature value is expressed as r _{(1) (2)} (−1 ≦ r _{(1) (3)} ≦ 1),. Correlation coefficient r _{(34) (35)} with the 35th feature quantity (−1 ≦ r _{(34) (35)} ≦ 1), and the correlation coefficient between the 35th feature quantity and the 35th feature quantity is represented by r _{(35) (35)} A matrix of 35 rows and 35 columns of Equation (3) of the following Equation 1 having (r _{(35) (35)} = 1) as each component is calculated as a correlation matrix R.
... Formula (3)

C2N: Correlation matrix storage means The correlation matrix storage means C2N has correlation coefficients r _{(1) (1)} , r _{(1) (2)} ,..., R _{(34) (35)} , r ₍₃₅₎ for each user. Correlation coefficient storage means C2N1 for storing ₍₃₅₎ and the user in association with each other, and storing the correlation matrix R for each user in association with the user. The correlation matrix storage unit C2N according to the first embodiment stores the correlation matrix R calculated by the correlation matrix calculation unit C2M and the user in association with each other. Specifically, when the correlation matrix R calculated for the user of “registration number p”, that is, the p-th user is R _p , the calculated “correlation matrix R _p ” and the user "Registration number p" is stored in association with each other.

FIG. 8 is an explanatory diagram of a registration start image of the first embodiment.
C2P: Registration Completion Image Display Unit The registration completion image display unit C2P performs registration as shown in FIG. 8 when the trial completion determination unit C2K determines that the trial for calculating and storing each feature amount has been completed. A registration completion image 105 for notifying completion is displayed on the front monitor FM. In FIG. 8, the registration completion image 105 of the first embodiment has a registration confirmation button 105a for allowing the user to confirm registration.
C2Q: Registration Confirmation Discriminating Unit The registration confirmation discriminating unit C2Q determines whether or not the user has confirmed registration by determining whether or not the registration confirmation button 105a has been selected. Note that the registration unit C2 of the first embodiment hides the registration completion image 105 when the user confirms registration. For this reason, the main image 101 is displayed again on the front monitor FM.

C3: Lock discriminating means The lock discriminating means (separation discriminating means, leaving seat discriminating means) C3, like the detection completion discriminating means C2H, has not detected all the pressure values of each button sensor SN1 of the sensor seat SN. (Whether or not it is zero), that is, whether or not the user has left the seat CH1 of the driver's seat CH makes the engine V (not shown) startable impossible. That is, it is determined whether or not to lock the engine start.
C4: Engine start lock means The engine start lock means C4 disables the start of the engine of the automobile V (locks the engine start) when it is determined that the user has left the driver's seat CH. The engine start lock means C4 of the first embodiment locks the engine start by controlling the engine so that fuel injection is not performed, for example.

C5: Authentication means The authentication means C5 includes a seating determination means C5A, a feature amount calculation means C5B, a distance calculation means C5C, a threshold value determination means C5D, a user authentication means C5E, a number counting means C5F, and a number determination means. Use of C5G, authentication display means C5H, and authentication confirmation determination means C5J, registered by a tester (subject, tester, subject, seated person, third party) seated in the driver's seat CH Authenticate whether or not In the case where there are MP registered users, the authentication unit C5 according to the first embodiment performs authentication in the order of “registration number p” from the first user to the MP user. If it is authenticated that the user is the user of the person (p = 1, 2,..., MP), the process ends without authenticating the remaining (MP-p) users.

C5A: Seating discriminating means The seating discriminating means C5A is, like the detection start discriminating means C2D, whether or not the sensor seat SN has started to detect pressure distribution, that is, the examiner is seated on the seat surface CH1 of the driver's seat CH. It is determined whether or not.
C5B: Feature amount calculating means The feature amount calculating means C5B, like the feature amount calculating means C2E, is based on the pressure values detected by the button sensors SN1 of the sensor sheet SN and includes 35 types of feature amounts (first (Feature amount, second feature amount,..., 35th feature amount) are calculated.

C5C: Distance calculation means The distance calculation means C5C is based on the correlation matrix R for each user stored in the correlation matrix storage means C2N and each feature quantity of the examiner calculated by the feature quantity calculation means C5B. , Calculate the Mahalanobis distance of the exhibitor. Here, the Mahalanobis distance is a comprehensive index (scale, statistical distance) of the multivariable values based on the correlation between multivariables, classification (character recognition, speech recognition, image recognition, etc.), prediction (Profit prediction, earthquake prediction, strength prediction, etc.) and diagnosis (disease diagnosis, fire alarm, industrial inspection, failure diagnosis, etc.). In particular, the MT system (MT method: Mahalanobis Taguchi Method) as a pattern recognition technology that introduces the concept of Mahalanobis distance is positioned as a major field in quality engineering, and covers a wide range of product planning, design, manufacturing, quality control, etc. It is used in the field. The Mahalanobis distance and MT system are described in, for example, Non-Patent Document 3 (“Judgment Technology for Automating Sensory Inspection”) and Non-Patent Document 4 (“Data Science for Environment and Quality”). It is described and publicly known.

The distance calculation means C5C according to the first embodiment first standardizes the calculated feature quantities of the presenter based on the feature quantities of the p-th user stored in the feature quantity storage means C2F. Specifically, i is an arbitrary natural number from 1 to 35, and the average of the i-th feature amount (MN i-th feature amount) of the p-th user is μ _ip (μ _ip = μ _1p , μ _2p , ..., and mu _35p), the standard deviation _{σ ip (σ ip = σ 1p} , σ 2p, ..., σ 35p) and then, each feature amount of the computed披験's _{v i (v i = v 1} , v ₂ ,..., V ₃₅ ), and each feature quantity of the _{normalized exhibitor} is u _ip (u _ip = u _1p , u _2p ,..., U _35p ), The feature quantity u _ip is calculated by the following equation (4).
u _ip = (v _i −μ _ip ) / σ _ip (4)

Then, a feature matrix which is a matrix in which 35-dimensional vectors of the normalized feature quantities u _ip of the examiner are represented in 35 rows and 1 column is represented by U _p, and a transposed matrix (1 row 35) of the feature matrix U _p is used. Column) is U _p ^T , the p-th user's correlation coefficient r _ij is r _ijp , the inverse matrix of the _p-th user's correlation matrix R _p is R _p ⁻¹ , and the p-th user's If the Mahalanobis distance披験's as authentication criteria and D _p, the square of the Mahalanobis distance D _p (D p ^₂₎ is calculated by the following equation (5) and the number 2 in the formula (6).
^{_{D p 2 = U p T (}} R p -1) U p ... (5)

... Formula (6)
Therefore, in the distance calculating means C5C of Example 1, and calculates the Mahalanobis distance _{D p} by the equation square root of the computed value _D ^{p 2} by a matrix equation shown in ^{_{(6) ((D p 2}} ) 1/2 = _Dp ).

C5D: threshold determination unit threshold determination means C5D is Mahalanobis distance D _p of the computed披験user to equal to or smaller than a predetermined threshold value S _D that by the distance calculating means C5C. In the threshold discriminating means C5D according to the first embodiment, the threshold _SD is set to “200” in advance.
C5E: user authentication means user authentication means (registrant authentication means) C5E, when the Mahalanobis distance D _p of披験individual is determined to be equal to or less than the threshold value S _D,披験person, the Mahalanobis distance D _p The user is authenticated as the p-th user associated with the correlation matrix R _p used in the calculation.

C5F: Number of people counting means Number of people counting means (authenticated number counting means) C5F counts the number of persons p who have authenticated whether or not the examiner is a registered user.
C5G: Number determination means Number determination means (authentication number determination means) C5G determines whether the number p counted by the number counting means C5F has reached a preset maximum value MP [person]. It is determined whether or not it is authenticated that the exhibitor is not a registered user of all the registered MP persons (determining whether or not the authentication is completed).
C5H: Authentication display means The authentication display means C5H has a success image display means C5H1 and a failure image display means C5H2, and displays an authentication result (that authentication has succeeded or authentication has failed).

FIG. 9 is an explanatory diagram of an image displaying the authentication result of the first embodiment, FIG. 9A is an explanatory diagram of a success image of the first embodiment, and FIG. 9B is an explanatory diagram of a failure image of the first embodiment.
C5H1: Success image display means The success image display means C5H1 is shown in FIG. 9A, when the exhibitor is authenticated by the user authentication means C5E. A success image 201 indicating that the authentication has succeeded because the user is a user of the. In FIG. 9A, the success image 201 according to the first embodiment includes a success confirmation button 201a for allowing the examiner to confirm that the authentication is successful.
C5H2: Failure image display means When the number determination means C5G determines that the authentication is completed, the failure image display means C5H2 is authenticated because the exhibitors are not all registered MP users as shown in FIG. 9B. Is displayed on the front monitor FM. In FIG. 9B, the failure image 202 of the first embodiment has a failure confirmation button 202a for allowing the examiner to confirm that the authentication has failed.

C5J: Authentication Confirmation Discriminating Unit The authentication confirmation discriminating unit C5J has a success confirmation discriminating unit C5J1 and a failure confirmation discriminating unit C5J2, and discriminates whether or not an authentication result has been confirmed by the examiner.
C5J1: Success confirmation determination means The success confirmation determination means C5J1 determines whether or not the presenter, that is, the pth user has confirmed the authentication success by determining whether or not the success confirmation button 201a is selected. Is determined.
C5J2: Failure confirmation determination means The failure confirmation determination means C5J2 determines whether or not the examiner has confirmed the authentication failure by determining whether or not the failure confirmation button 202a has been selected.
C6: Engine start lock release means The engine start lock release means C6 releases the engine start lock of the vehicle V when the user authentication means C5E authenticates that the test person is the p-th user.

C7: Registration Update Unit Registration update unit (correlation coefficient re-learning unit, re-learning unit) C7 includes feature amount update unit C7A, unit space regeneration unit C7B, correlation matrix recalculation unit C7C, and correlation matrix update unit. C7D and update the user registration (re-learn the correlation matrix R _p for each user).
C7A: Feature amount update unit The feature amount update unit C7A is configured to calculate each feature amount calculated by the feature amount calculation unit C5B when the tester is authenticated as the p-th user by the user authentication unit C5E. Update memory. The feature amount update unit C7A according to the first exemplary embodiment sets the feature amount of the p-th user stored in the feature amount storage unit C2F for the longest time, that is, the feature amount of the MN set in the first order in the storage order. The feature amount of the eye, the second feature amount,..., Calculated by the feature amount calculation means C5B instead of the first feature amount when the 35th feature amount of the MN set is used. By storing each feature quantity, each feature quantity stored in the feature quantity storage means C2F is updated.

C7B: Unit space regenerating means The unit space regenerating means (reference space regenerating means) C7B, as with the unit space generating means C2L, is the feature quantity (MN pieces) of MN sets updated by the feature quantity updating means C7A. The unit space of the pth user is regenerated based on the combination of the first feature value, the MN second feature value,..., The MN 35th feature value).
C7C: Correlation matrix recalculation means Correlation matrix recalculation means (correlation coefficient matrix recalculation means) C7C is provided in the unit space regenerated by the unit space regeneration means C7B, similarly to the correlation matrix calculation means C2M. Based on this, the correlation matrix R is recalculated.
C7D: correlation matrix updating means correlation matrix updating means (correlation coefficient updating means) C7D, the place of the correlation matrix R _p of the user of the p person stored in the correlation matrix storage unit C2N, the correlation matrix recalculation means by storing the correlation matrix R _p calculated by the C7C, updating the correlation matrix R _p, which is stored in the correlation matrix storage means C2N.

(Description of Flowchart of Example 1)
Next, the process flow of each program of the control unit C according to the first embodiment will be described with reference to flowcharts. As for the process of displaying the main image 101 (see FIG. 5), only the main image 101 is displayed on the front monitor FM when the vehicle V starts the engine. Since only the engine start is locked when the sensor SN detects the absence of the user, illustration and detailed description thereof are omitted.

(Description of registration process of control unit C of embodiment 1)
FIG. 10 is a flowchart of the registration process according to the first embodiment.
The processing of each ST (step) in the flowchart of FIG. 10 is performed according to a program stored in the ROM or the like of the control unit C. In addition, this process is executed by multitasking in parallel with other various processes of the control unit C.

The flowchart shown in FIG. 10 is started when the personal authentication program AP1 of the control unit C is activated.
In ST1 of FIG. 10, it is determined whether or not the user registration is started by determining whether or not the registration start button 101a is selected from the main image 101. If yes (Y), the process proceeds to ST2, and if no (N), ST1 is repeated.
In ST2, the following processes (1) and (2) are executed, and the process proceeds to ST3.
(1) The registration start image 102 (see FIG. 6) is displayed on the front monitor FM.
(2) Set 1 to the number of trials n (n = 1)
In ST3, the detection start guide image 103 (see FIG. 7A) is displayed on the guide display unit 102a. Then, the process proceeds to ST4.

In ST4, it is determined whether or not the user is seated in the driver's seat CH by determining whether or not each pressure value has changed after each button sensor SN1 of the sensor seat SN detects the pressure value. To do. That is, it is determined whether or not each button sensor SN1 of the pressure detection sensor SN detects a pressure value (outputs a pressure distribution) from any one of the 360 square detection regions. If yes (Y), the process proceeds to ST5, and, if no (N), ST4 is repeated.
In ST5, 35 types of feature amounts are calculated and stored from the pressure values of the 360 button sensors SN1. Then, the process proceeds to ST6.
In ST6, it is determined whether or not the number of trials n is greater than or equal to the maximum value MN [times] (n ≧ MN). If no (N), the process moves to ST7, and if yes (Y), the process moves to ST9.
In ST7, the following processes (1) and (2) are executed, and the process proceeds to ST8.
(1) The detection completion guide image 104 (see FIG. 7B) is displayed on the guide display unit 102a.
(2) Add +1 to the number of trials n (n = n + 1).

In ST8, it is determined whether or not the user has left the driver's seat CH by determining whether or not each button sensor SN1 of the sensor seat SN no longer detects each pressure value. If yes (Y), the process returns to ST3, and if no (N), ST8 is repeated.
In ST9, the following processes (1) to (4) are executed, and the process proceeds to ST10.
(1) A unit space for the user is generated by normalizing each feature amount to an average of 0 and a standard deviation of 1.
(2) A 35-by-35 correlation matrix R of Equation (3) of Equation 1 is calculated from the generated unit space, using the correlation coefficient r _ij calculated by Equation (2) as each component.
(3) The user and the calculated correlation matrix R are stored in association with each other.
(4) The registration completion image 105 (see FIG. 8) is displayed on the front monitor FM.

In ST10, it is determined whether or not the user has confirmed registration by determining whether or not the registration confirmation button 105a has been selected. If yes (Y), the process proceeds to ST11, and, if no (N), ST10 is repeated.
In ST11, the main image 101 is displayed again on the front monitor FM by hiding the registration completion image 105. Then, the process returns to ST1.

(Description of Authentication Process of Control Unit C of Example 1)
FIG. 11 is a flowchart of the authentication process according to the first embodiment.
Each ST (step) process in the flowchart of FIG. 11 is performed according to a program stored in the ROM or the like of the control unit C. In addition, this process is executed by multitasking in parallel with other various processes of the control unit C.

The flowchart shown in FIG. 11 is started when the personal authentication program AP1 of the control unit C is activated.
In ST101 of FIG. 11, whether or not the examiner is seated in the driver's seat CH by determining whether or not each pressure value has stopped changing after each button sensor SN1 of the sensor seat SN detects each pressure value. Is determined. If yes (Y), the process transfers to ST102, and, if no (N), ST101 is repeated.
In ST102, the following processes (1) and (2) are executed, and the process proceeds to ST103.
(1) 35 types of feature values are calculated from the pressure values of the 360 button sensors SN1.
(2) Set 1 to the number of people p (p = 1).

In ST103, the feature matrixes U _p ^T and U _p (35 rows and 1 column, 1 row and 35 columns) based on the calculated feature amounts and the correlation matrix of the p-th user according to the equations (4) to (6) Based on R _p (35 rows and 35 columns), the Mahalanobis distance D _{p of the} examiner as the authentication standard of the _p-th user is calculated. Then, the process proceeds to ST104.
In ST 104, the Mahalanobis distance _{D p} is equal to or smaller than a threshold _{S D (S D = 200)} (D ≦ 200). If yes (Y), the process proceeds to ST105, and, if no (N), the process proceeds to ST108.
In ST105, the following processes (1) and (2) are executed, and the process proceeds to ST106.
(1) The success image 201 (see FIG. 9A) is displayed on the front monitor FM.
(2) Release the engine start lock.

In ST106, a registration update process shown in FIG. Then, the process proceeds to ST107.
In ST107, by determining whether or not the success confirmation button 201a has been selected, it is determined whether or not the presenter, that is, the pth user has confirmed the success of the authentication. If yes (Y), the process proceeds to ST112, and, if no (N), ST107 is repeated.
In ST108, it is determined whether or not the number of persons p is equal to or greater than the maximum value MP [person] (p ≧ MP). If no (N), the process moves to ST109, and if yes (Y), the process moves to ST110.

In ST109, +1 is added to the number of people p (p = p + 1). Then, the process returns to ST103.
In ST110, the failure image 202 (see FIG. 9B) is displayed on the front monitor FM. Then, the process proceeds to ST111.
In ST111, by determining whether or not the failure confirmation button 202a has been selected, it is determined whether or not the examiner has confirmed authentication failure. If yes (Y), the process proceeds to ST112, and, if no (N), ST111 is repeated.
In ST112, the images 201 and 202 are not displayed. Then, the process returns to ST101.

(Description of Flowchart of Registration Update Processing of Control Unit C of Example 1)
FIG. 12 is a flowchart of the registration update process of the first embodiment, and is an explanatory diagram of the subroutine of ST106 of FIG.
In ST201 of FIG. 12, each feature quantity of the first set with the earliest storage order is updated to each feature quantity calculated in ST102 of FIG.
In ST202, each feature amount is normalized to an average of 0 and a standard deviation of 1 to regenerate the user unit space. Then, the process proceeds to ST203.
In ST203, the following processes (1) and (2) are executed, the registration update process is terminated, and the process returns to ST106 in FIG.
(1) _Recalculate the 35 _× 35 correlation matrix R _p of Equation (3) of Equation 1 _using the correlation coefficient r _ijp calculated by Equation (2) from the regenerated unit space as each component.
(2) p person of the user and re-computed correlation matrix in association with R _p storing updated.

(Operation of Example 1)
In the personal authentication system S of the first embodiment having the above-described configuration, the correlation matrix R _{1 to} R _MP of each user of the vehicle V is registered by the registration process (see ST1 to ST11 in FIG. 10). When the presenter sits in the driver's seat CH, the authentication process (see ST101 to ST112 in FIG. 11) for authenticating whether the presenter is one of the users is executed.

In the authentication process of the first embodiment, after the examiner is seated on the seat surface CH1 of the driver's seat CH and each pressure sensor no longer changes after each button sensor SN1 of the sensor seat SN detects each pressure value, Thirty-five types of feature values are calculated from the pressure values of the 360 button sensors SN1. Then, the Mahalanobis distance D _{p of the} examiner calculated based on the feature matrix U _p based on the calculated feature amount and the correlation matrix R _p of each user stored in advance (Expressions (4) to (6) )) Is less than or equal to the threshold S _D (S _D = 200), it is authenticated that the exhibitor is a user.
Therefore, the personal authentication system S according to the first embodiment can perform personal authentication using the MT system in quality engineering.

  In Patent Document 1, the pressure distribution data (measurement information) based on the pressure sensor signal and the user's pre-stored pressure distribution data (storage information, registration information) match within a predetermined error range. It is determined whether or not to do. For this reason, for example, when the seated posture is different from that at the time of registration, there is a possibility that the pressure distribution data of the person does not match within the predetermined error range. Also, for example, the physical distribution such as “weight”, “seat surface area”, “seat surface shape”, and the like may have matched the pressure distribution data of another person close to the user within the predetermined error range. .

Moreover, in the said patent document 2, the outer contour line and grounding point of a shoe sole based on the signal of a shoe sole pressure sensor, ie, a shoe sole pressure pattern (measurement information), and the user's shoe sole pressure pattern ( It is determined whether or not (stored information, registered information) matches. For this reason, as in Patent Document 1, for example, when the posture on the shoe sole pressure sensor is different from that at the time of registration, there is a possibility that the shoe sole pressure pattern of the person does not match. In addition, for example, there is a possibility that other person's shoe sole pressure patterns close to the person have physical characteristics such as “weight”, “shoe size”, and “shoe sole area”.
As a result, as in Patent Documents 1 and 2, when personal authentication is performed using the pressure sensor measurement information (pressure distribution data, shoe sole pressure pattern) as input parameters, There was a problem that there was a high possibility that the person would be mistaken for the person.

On the other hand, in the first embodiment, each feature quantity that specifies the relationship between the pressure values detected by each button sensor SN1, that is, the value of an individual physical feature is used as an input parameter. Further, in the first embodiment, a correlation coefficient r _ijp (stored information, registered information) stored in advance that specifies the correlation between two different feature quantities, that is, a value based on an individual difference in correlation between physical features. We are doing a personal authentication by calculating the Mahalanobis distance D _p from.

For this reason, for example, even when the posture of the person sitting on the sensor sheet SN is different from that at the time of registration, the correlation coefficient r _ijp is calculated based on the change of each feature amount according to the change of the pressure distribution due to the difference of the posture. and for which the value of the Mahalanobis distance D _p of the person, which is calculated from the correlation coefficient of the principal r _ijp is reduced. Also, for example, even if the physical features such as “weight”, “sitting surface area”, and “sitting surface shape” are close to the person, the pressure distribution depends on other physical characteristics such as “pelvic skeleton”. Since there is an individual difference and there is also an individual difference in the correlation between the change in “weight” and the change in “the area of the seating surface”, the Mahalanobis distance of another person calculated from the correlation coefficient r _{ijp of} the person the value of D _p increases.
As a result, the personal authentication system S according to the first embodiment recognizes the person himself / herself and is not the person himself / herself even when the pressure distribution at the time of measurement changes from the time of registration as compared with the configurations of the patent documents 1 and 2. It is possible to increase the accuracy of personal authentication that is recognized.

  Moreover, in the said nonpatent literature 1, the output value calculated based on the pressure distribution data (measurement information) based on the signal of a pressure sensor, and the joint load (memory information, registration information) memorize | stored beforehand of the support vector machine. The personal authentication is done by. That is, as in Patent Documents 1 and 2, personal authentication is performed using measurement information (pressure distribution data, shoe sole pressure pattern) of the pressure sensor as it is as an input parameter. As a result, if the learning data (training sample) is not sufficient, the person's pressure distribution data changes due to the difference in the seated posture, and the person is misidentified as not being the person, or the "weight" or "sitting area" or "sitting There is a possibility that the physical characteristics such as “the shape of the surface” may be misidentified as the person in the pressure distribution data of others close to the person.

On the other hand, in the first embodiment, the correlation coefficient r _ijp is calculated as a value based on the individual difference of the correlation between the physical features, and even if the learning data (training sample) is not sufficient, the person and others It is easy to obtain individual differences in the value of the correlation coefficient r _ijp . Therefore, in the first embodiment, even if the learning amount is small, the value of the Mahalanobis distance D _p of the principal calculated from the correlation coefficient r _{ijp of the} principal _tends to be small and the value of the Mahalanobis distance D _p of others is likely to be large. Yes. As a result, the personal authentication system S according to the first embodiment increases the accuracy of personal authentication that recognizes the person as the principal and recognizes that the other person is not the person even with a small learning amount, as compared with the configuration of the non-patent document 1. Can do.

(Experimental example)
Here, in order to examine whether the individual authentication system S of the first embodiment using the MT system can obtain a high identification rate by executing the authentication process, five subjects (exhibitor and user) On the other hand, an experiment was conducted in which sitting and leaving were repeated while leaning on the back CH2 of the driver's seat CH and the legs naturally lowered.
(Experimental conditions)
In the experimental example, for the registration process, five subjects (A, B, C, D, E) were seated on and off from the seat surface CH1 of the driver's seat CH twice (two sets). ) by trial is made to register the feature amount and the correlation matrix _{R p} of each subject in turn (MP = 5, MN = 2 ). For each subject, seating / seating on the seat CH1 is performed twice (two sets), and the pressure distribution is 300 frames in total, 30 frames per second for 10 seconds. It was allowed to output to the sensor sheet SN registered the feature quantity and the correlation matrix R _p. That is, each feature amount for 5 people is calculated by 600 sets (2 × 300 = 600 [pieces]) by 2 sets of sitting and leaving each subject, and the correlation matrix for 5 people is calculated (stored). It was.

In the authentication process, the processes of ST104 to ST107, ST110 to ST112 are omitted, and only the processes of ST101 to ST103, ST108, and ST109 are executed. That is, in the experimental example, determination by the threshold _{S D} (personal authentication) and the result without executing the processing for displaying the Mahalanobis distance _D p _{(D 1} of all _{five, D 2, D 3, D} 4, D ₅ ) was calculated (stored). In addition, for each subject, seating / seating on the seat surface CH1 is performed once (one set), and a pressure distribution of 300 frames in total for 30 seconds per second for 10 seconds in one set. and computing the Mahalanobis distance D _p to thereby output the sensor sheet SN. That is, 300 pieces of each of the five types of Mahalanobis distances D _{1 to} D ₅ were calculated (stored) by one set of sitting and leaving each subject.
In the experimental example, “Conform-Light” manufactured by Nitta Corporation was used as the sensor sheet SN.

(Experimental result)
FIG. 13 is a diagram for explaining the operation of the first embodiment. FIG. 13A is a semilogarithmic graph of the authentication result in which the vertical axis is a common logarithm of Mahalanobis distance and the horizontal axis is an equally divided frame for each of five subjects. FIG. 13B is a graph of the authentication result of each subject when subject B is the user when subject A is the user.
In the experimental example, as shown in FIG. 13A, when the user is the subject A, that is, when the Mahalanobis distance D ₁ is calculated from the correlation matrix R ₁ of the subject A, the subject A himself / herself has the Mahalanobis distance D _It can be seen that ₁ is all 200 or less (D ₁ ≦ S _D ). That is, it can be seen that the principal rejection rate is 0 [%] (the principal acceptance rate is 100 [%]). It can also be seen that for the other subjects B to E, the Mahalanobis distance D ₁ exceeds 200 (D ₁ > S _D ). That is, it can be seen that the other-person tolerance rate is 0 [%] (other-person rejection rate is 100 [%]).

As shown in FIG. 13B, when the user is the subject B and the Mahalanobis distance D ₂ is calculated using the correlation matrix R ₂ of the subject B, the rejection rate of the person is the same as when the user is the subject A. Is 0 [%] (Personal allowance is 100 [%]) and the Others allowance is 0 [%] (Others rejection is 100 [%]). In the case where the users are set as subjects C to E, the rejection rate of the person is 0 [%] (the allowable rate of the person is 100 [%]) and the other person is allowed as in the case where the users are the subjects A and B. It was confirmed that the rate was 0 [%] (others rejection rate was 100 [%]).
As a result, the personal authentication system S according to the first embodiment does not identify the person himself / herself due to a slight difference in the position or posture of the seating on the sensor seat SN or a slight change in the posture of the sensor seat SN while sitting. It has been confirmed that it is possible to perform personal authentication without misidentifying the person or misidentifying another person as the person.
In addition, the personal authentication system S according to the first embodiment can execute personal authentication with a high identification rate by executing only two sets of trials for sitting and leaving the seat CH1 of the driver's seat CH. It was confirmed that

Further, in the personal authentication system S of the first embodiment having the above-described configuration, when the tester is authenticated as the p-th user by the authentication process, the registration update process performs registration as registration information. the registration updating process for updating the correlation matrix _{R p,} etc. (see ST201~ST203 in FIG. 12) is executed.
As a result, the personal authentication system S according to the first embodiment can authenticate whether or not the exhibitor is a user by using the correlation matrix R _p reflecting the feature amount based on the latest pressure distribution of the user. it can. Therefore, even when the physical characteristics (weight, etc.) of the user change over time from the time of registration, it is possible to reduce misidentification of the person as not being the person by repeating the registration update process.

Next, the second embodiment of the present invention will be described. In the description of the second embodiment, the same reference numerals are given to the components corresponding to the components of the first embodiment, and the detailed description thereof will be given. Omitted.
The second embodiment is different from the first embodiment in the following points, but is configured in the same manner as the first embodiment in other points.

FIG. 14 is an overall explanatory diagram of the personal authentication system of the second embodiment, and is an explanatory diagram corresponding to FIG. 1B of the first embodiment.
In FIG. 14, the personal authentication system S ′ according to the second embodiment of the present invention is managed by a user (examinee, registrant, room entrant, person permitted to enter, room manager or responsible person). It is provided in the room entrance management device (personal authentication device) NK of the automatic door AD of the room. The automatic door AD according to the second embodiment includes a single sliding door AD1 that is slidable in the left-right direction, and a door pocket AD2 that houses the single sliding door AD1 slid rightward. That is, the one-way door AD1 of Example 2 is configured to be slidable between a closed position that closes the entrance of the management room and an open position that is accommodated in the door pocket AD2 and opens the entrance.
As a result, in the second embodiment, it is natural for the user to stand upright in front of the one-side sliding door AD1 disposed at the closed position.

Further, the sensor sheet SN of Example 2 is laid in front of the closed position on the floor surface (corridor) FL in front of the automatic door AD. The sensor sheet SN according to the second embodiment outputs the pressure distribution on the sole of the user who stands upright. The sensor sheet SN according to the second embodiment is different from the sensor sheet SN according to the first embodiment in that the sensor sheet SN is configured to be wirelessly connectable to the room management device NK and outputs a pressure distribution. About the structure, it is comprised similarly to the said sensor sheet SN.
In addition, the room entry management device NK of the second embodiment is configured by a so-called computer device, and is configured by a control unit or the like (not shown).

(Description of the control unit of the room entry management device NK of the second embodiment)
FIG. 15 is an explanatory diagram showing the function of the room entry management device of the second embodiment in a block diagram (functional block diagram), and is an explanatory diagram corresponding to FIG. 3 of the first embodiment.
In FIG. 15, the control unit of the room management device (personal authentication device) NK according to the second embodiment includes the main image display unit C1, the registration unit C2, and the registration update unit C7 according to the first embodiment. . Further, the control unit replaces the lock determination means C3, the engine start lock means C4, the authentication means C5, and the engine start lock release means C6 with a lock determination means C3 ′, an automatic door lock. Means C4 ', authentication means C5', and automatic door lock releasing means C6 '. Further, automatic door control means C8 is newly added to the control unit.
In addition, the main image display means C1 of Example 2 displays the main image 101 (refer FIG. 5) regarding control of automatic door AD on the front monitor FM of the entrance management apparatus NK. Further, in the feature amount calculation unit C2E of the registration unit C2 of the second embodiment, the following features (1 ′) to (35 ′) are preset in place of the (1) to (35) of the first embodiment. ing.

(1 ') Area of the soles of the user's feet (2') Average of each pressure value in the button sensor having a pressure value of 1 or more (3 ') Each pressure value in the button sensor having a pressure value of 1 or more Variance (4 ') Variance of each pressure value in a button sensor having a pressure value of 1-50 (5') Variance of each pressure value in a button sensor having a pressure value of 51-100 (6 ') 101-150 Dispersion of each pressure value in a button sensor having a pressure value (7 ') Dispersion of each pressure value in a button sensor having a pressure value of 151-200 (8') In a button sensor having a pressure value of 201-255 (9 ') Number of button sensors having a pressure value of 1 or more (10') Number of button sensors having a pressure value of 1-50

(11 ′) Number of button sensors having pressure values of 51 to 100 (12 ′) Number of button sensors having pressure values of 101 to 150 (13 ′) Number of button sensors having pressure values of 151 to 200 (14 ') Number of button sensors having pressure values of 201-255 (15') Degree of extension of both feet of the user (16 ') Circularity of the feet of the user (17') Maximum pressure of both feet of the user Total area (18 ') total pressure (total value of each pressure value)
(19 ') Perimeter of the user's soles

(20 ') The area of the toes on both soles of the user (21') The average of each pressure value of only the button sensor having one or more pressure values on the toes on both soles of the user (22 ') Dispersion of each pressure value of only the button sensor having a pressure value of 1 or more at the toe of the user (23 ') Area of the arch of the user's both soles (24') One or more pressure value of the arch of the user's both soles Average of each pressure value of only the button sensor having (25 ') Variance of each pressure value of only the button sensor having one or more pressure values on the arch of the user's soles

(26 ') The area of the heel of both feet of the user (27') The average of each pressure value of only the button sensor having one or more pressure values on the heel of the feet of the user (28 ') Both feet of the user Dispersion of each pressure value of only the button sensor having a pressure value of 1 or more at the heel (29 ') Distance between the center of the area and the center of load (30') Maximum pressure of the whole (maximum value of each pressure value)
(31 ′) Secondary moment in a button sensor having a pressure value of 1 to 50 (32 ′) Secondary pressure in a button sensor having a pressure value of 51 to 100 (33 ′) having a pressure value of 101 to 150 Secondary moment in the button sensor having a pressure value of secondary moment (35 ') 201-255 in the button sensor having a pressure value of secondary moment (34') 151-200 in the button sensor

C3 ': Lock determination means The lock determination means (separation determination means) C3' determines whether or not to lock the automatic door AD. The lock discriminating means C3 ′ according to the second embodiment is an automatic door when the one-side sliding door AD1 of the automatic door AD is disposed at the closed position and each button sensor SN1 of the sensor seat SN does not detect each pressure value. It is determined that AD lock is executed.
C4 ': Automatic door locking means When the automatic door locking means C4' is determined to lock the automatic door AD, the automatic door AD makes the sliding door AD1 non-slidable. That is, the opening / closing of the automatic door AD is locked.

C5 ': Authentication means The authentication means C5' is an upright discrimination means C5A 'instead of the seating discrimination means C5A, the authentication display means C5H, and the authentication confirmation discrimination means C5J, compared to the authentication means C5 of the first embodiment. And an authentication display means C5H 'and an authentication confirmation determination means C5J'. The authentication means C5 ′ authenticates whether or not an upright examiner (subject, examiner, subject, upright, third party) is a registered user. Note that the feature amount calculation unit C5B of the second embodiment calculates the feature amounts of the subjects (1 ′) to (35 ′) in the same manner as the feature amount calculation unit C2E.
C5A ': Upright discriminating means The upright discriminating means C5A, like the detection start discriminating means C2D, determines whether or not the sensor sheet SN has started to detect the pressure distribution, that is, the examiner applies the sensor sheet SN on the floor surface FL. Determine if you are riding and standing upright.

C5H ': Authentication display means The authentication display means C5H' displays an authentication result (that authentication has been successful, authentication has failed). When the user authentication means C5E authenticates that the test person is the p-th user, the authentication display means C5H ′ of the second embodiment sounds a success notification sound indicating that the authentication is successful, and When the tester is authenticated by the user authentication means C5E as the p-th user, the authentication result is displayed by sounding a failure notification sound indicating that the authentication has failed.
C5J ′: Authentication Confirmation Discriminating Unit The authentication confirmation discriminating unit C5J ′ discriminates whether or not the test result has been confirmed by the examiner. The authentication confirmation determination means C5J ′ of the second embodiment determines whether or not the tester has confirmed the authentication failure by determining whether or not each button sensor SN1 of the sensor sheet SN no longer detects each pressure value. Is determined.

C6 ': Automatic door lock releasing means Automatic door lock releasing means C6' releases the lock on opening and closing of the automatic door AD when the user authentication means C5E authenticates that the subject is the pth user. .
C8: Automatic door control means The automatic door control means C8 is located between the closed position and the open position of the single door AD1 of the automatic door AD when the automatic door AD is unlocked by the automatic door lock releasing means C6 ′. Control slide movement.

(Explanation of flowchart of embodiment 2)
Next, the processing flow of each program of the room entrance management device NK according to the second embodiment will be described with reference to flowcharts. As for the registration process of the second embodiment, “sitting on the driver's seat CH” in the registration process of the first embodiment (see FIG. 10) is “upright on the sensor seat SN”, and “separated from the driver's seat CH”. Is simply “separated from the sensor sheet SN”. Further, the registration update process of the second embodiment is the same as the registration update process (see FIG. 12) of the first embodiment, and thus illustration and detailed description thereof are omitted.

(Description of authentication process of room entry management device NK of embodiment 2)
FIG. 16 is a flowchart of the authentication process performed by the room entrance management apparatus according to the second embodiment, and is an explanatory diagram corresponding to FIG. 11 according to the first embodiment.
In FIG. 16, the authentication process of the second embodiment is performed by replacing ST101, ST105, ST110, and ST111 of the authentication process (see FIG. 11) of the first embodiment with ST101 ′, ST105 ′, ST110 ′, and ST111 ′. ST107 and ST112 of Example 1 are omitted. Further, since the authentication process is the same as that of the first embodiment except for returning to ST101 ′ after executing ST106, detailed description thereof is omitted.

In ST101 ′ of FIG. 16, the examiner stands upright on the sensor sheet SN by determining whether or not each pressure value has changed after each button sensor SN1 of the sensor sheet SN detects the pressure value. It is determined whether or not. If yes (Y), the process transfers to ST102, and, if no (N), ST101 'is repeated.
In ST105 ′, the following processes (1) to (3) are executed, and the process proceeds to ST106.
(1) Sound success notification sound.
(2) Release the open / close lock of the automatic door AD.
(3) Open and close the automatic door AD.
In ST110 ', a failure notification sound is emitted. Then, the process returns to ST101 ′.
In ST111 ′, it is determined whether or not the subject has left the sensor sheet SN. If yes (Y), the process returns to ST101 ', and if no (N), the process returns to ST110'.

(Operation of Example 2)
In the personal authentication system S ′ according to the second embodiment having the above-described configuration, the correlation matrixes R _{1 to} R _{MP of MP} users are registered by the registration process (see ST1 to ST13 in FIG. 10). When the exhibitor stands upright on the sensor sheet SN, the authentication process (see ST101 ′ to ST111 ′ in FIG. 16) for authenticating whether the exhibitor is an MP user is executed.

In the authentication processing in Example 2, similarly to the authentication processing in Example 1,披験's upright riding sensor sheet SN on the floor surface FL, computed披験's Mahalanobis distance D _p is the threshold value S _{If it is D} (S _D = 200) or less, it is authenticated that the exhibitor is a user.
Therefore, the personal authentication system S ′ according to the second embodiment can perform personal authentication using the MT system in the same manner as the personal authentication system S according to the first embodiment.

In addition, as for the personal authentication system S ′ of the second embodiment, as in the first embodiment, it was tested whether a high identification rate could be obtained by executing the authentication process. Specifically, it is assumed that the personal authentication system S ′ is used in a factory or the like, and 20 subjects (exhibitors and users) wear electrostatic shoes as an example of the same type of internal wear. In this state, the experiment was repeated on the sensor sheet SN. As a result, it was confirmed that for all 20 subjects, the rejection rate of the principal was 0 [%] (acceptance rate of the principal was 100 [%]) and the acceptable rate of others was 0 [%] (rejection rate of others was 100 [%]). .
In addition, the personal authentication system S ′ according to the second embodiment has the same effects as the personal authentication system S according to the first embodiment.

(Example of change)
As mentioned above, although the Example of this invention was explained in full detail, this invention is not limited to the said Example, A various change is performed within the range of the summary of this invention described in the claim. It is possible. Modification examples (H01) to (H016) of the present invention are exemplified below.
(H01) In the sensor sheet SN of the embodiment, 360 button sensors SN1 are arranged in a detection area of 20 squares in the vertical direction, 18 squares in the horizontal direction (cell), and 360 squares in total (20 × 18 = 360 [squares]). However, the present invention is not limited to this, and the number of button sensors SN1 can be increased or decreased to 2 or more, 359 or less, or 361 or more. In this case, for example, when two the number of buttons sensor SN1, which may or become the feature quantity for calculating the Mahalanobis distance D _p is impossible operation, inaccurate feature quantity or is calculated, It is necessary to reduce the number of button sensors SN1 according to the contents of each feature required by the personal authentication systems S and S ′, the number of types, the rejection rate [%] of the person, and the allowable rate [%] of others.

(H02) In the above-described embodiment, the sensor sheet SN is laid on the seating surface CH1 and the floor surface FL. However, the present invention is not limited to this. For example, the sensor sheet SN can be embedded in the seating surface CH1 or the floor surface FL. In this case, the button sensor (pressure-sensitive conductive ink) SN1 and the electrode SN2 are difficult to contact and may not be able to detect the pressure value. It is also possible to apply a member.
(H03) In the first embodiment, the sensor seat SN is laid only on the seat surface CH1 of the driver's seat CH on which the user and the subject are seated. However, the present invention is not limited to this. It is also possible to output the pressure distribution by arranging it on a seat belt or the like. Further, in order to perform personal authentication with high accuracy, it is preferable to lay the sensor seat SN on the seat surface CH1, but for example, the sensor seat SN of the seat surface CH1 is omitted and only the backrest CH2, armrest, seat belt, etc. It is also possible to output each pressure distribution by arranging a sheet.

(H04) In the above-described embodiment, 35 types of feature values are calculated based on each pressure value detected by each button sensor SN1, but the present invention is not limited to this. For example, the personal authentication systems S and S ′ require it. It is also possible to reduce the number of feature amounts from 2 to 34 types according to the rejection rate [%] of the principal and the allowable rate [%] of others. In the first embodiment, when only three types of feature values are calculated and the Mahalanobis distance D _p is calculated from the 1 × 3 feature value matrix U _p and the 3 × 3 correlation matrix R _p , (See ST103 in FIG. 11), the permissible person ratio [%] is approximately 60%, and only 5 types of feature quantities are calculated, and a 1 × 5 feature quantity matrix _Up and a 5 × 5 correlation matrix R are calculated. If the Mahalanobis distance D _p by the _p is computed, it can be 70% or more of the person permitted ratio [%] has been confirmed by experiments. It is also possible to increase the number of feature quantities to 36 or more. In addition, in the said Example 1, when it is set as 36 types or more, it is also possible to add the following features (36)-(43), for example.
(36) Number of button sensors having a pressure value equal to or greater than 50% of the left-side maximum pressure value adjacent to the button sensor having a pressure value of 0 on the left half of the user (37) The pressure value on the right half of the user is Number of button sensors having a pressure value equal to or greater than 50% of the right-side maximum pressure value adjacent to the zero button sensor (38) Distance from the rear end to the center of the user's buttocks (the rear end of the crack) (39) Distance from rear end to user's crotch (front end of crack) (40) Distance from left end to right end of user's left leg (41) Distance from left end to left end of user's right leg (42) Left maximum Number of button sensors adjacent to the top, bottom, left, and right of a button sensor having a pressure value of 75% or more of the pressure value (perimeter in squares surrounding the detection area)
(43) Number of button sensors adjacent to the upper, lower, left, and right sides of a button sensor having a pressure value of 75% or more of the maximum right pressure value

(H05) In the above embodiment, “registration number p”, “each feature amount of MN set”, and “correlation matrix R _p (correlation count r _ijp )” are stored in association with each other as user registration information. The registered information is not limited to this, and for example, personal information such as the user's name and face photo can be stored in association with each other.
(H06) In the above embodiment, the correlation matrix R is calculated by normalizing each of the 35 types of feature quantities so that the average is 0 and the standard deviation is 1. However, the correlation matrix R is calculated without normalization. It is also possible to do. In this case, the Mahalanobis distance D _p is calculated without reference by mean mu _ip and standard deviation sigma _ip of a user feature value of the computed subject. That is, equation (5) by omitting the formula (4), (6) the Mahalanobis distance _{D p} only is calculated.
(H07) In the embodiment, when the p-th user is authenticated by the registration update process (see FIG. 12), the correlation matrix R _p of the _p-th user is updated. However, the present invention is not limited to this. For example, update only once per multiple authentications, or update only when a preset update period (eg, 3 days, 1 week, 1 month, etc.) has passed since the previous authentication. It is also possible to do.
(H08) As in the above embodiment, it is preferable to update the correlation matrix R _p of the pth user authenticated by the registration update process (see FIG. 12), but this can be omitted. .

(H09) Each of the 35 types of features of the embodiment is not limited to this, and can be replaced with other features. For example, in the embodiment 1, the features of (1) to (35) It is also possible to replace the features (36) to (45). In this case, the features (1), (2), (11), (16), (32), and (33) are considered to have a particularly large proportion of contribution to improving the accuracy of personal authentication. It is preferable to leave it alone.
(H010) In the above embodiment, although the threshold value is set S _D of each user with respect to the Mahalanobis distance D _p of the subject 200 is not limited to this, personal authentication system S, the false rejection rate that S 'requests% ] Or other person's allowable rate [%].

(H011) In the first embodiment, the authentication result is confirmed by the confirmation buttons 201a and 202a of the success image 201 (see FIG. 9A) and the failure image 202 (see FIG. 9B). It can be omitted. In this case, for example, the images 201 and 202 can be automatically hidden after being displayed for a certain time (for example, 5 seconds).
(H012) In the above embodiment, the failure image 202 is displayed at the time of authentication failure or the failure notification sound is sounded. However, the present invention is not limited to this. For example, for crime prevention, an alarm (siren etc.) is sounded, It is also possible to report to the police, security company or security room.

(H013) In the first embodiment, when it is determined that a person other than the user is seated on the seat surface CH1 of the driver's seat CH, the function of the automobile V is completely limited. It is also possible to limit a part of the functions of the automobile V. For example, when it is determined that the person is another person, it is possible to control the engine to operate only at a low speed. In this case, a person such as a rescuer must drive to a predetermined position in a situation where the automobile V must be left in an earthquake, etc., or when driving to a hotel, pass the key to the staff and drive to the parking lot. Is possible. Further, the application range of the personal authentication system S of the first embodiment to the vehicle V is not limited to the limitation of the function of the vehicle V. For example, the seat position or the steering position registered in advance by the user, It is also possible to apply air-conditioning, playback of music CDs and car radios, or so-called customization. In addition, for example, it is possible to check which user the driver is when checking the accident status with a drive recorder or the like.

(H014) In the above-described embodiment, the personal authentication systems S and S ′ are used to unlock the engine start of the automobile V and to manage the entrance of the management room. , S ′ is not limited to this. For example, for personal authentication using a chair as in the first embodiment, automatic login of a personal computer (PC) at a company or home, attendance management at the company, At-home management at home, grasping the current location of patients and elderly people in hospitals and nursing homes, customer management at members (restaurants with seats, barber shops, beauty salons, etc.), entrance examinations, It can also be used to prevent taking a replacement for a qualification test.
In (H015) Example, the registration method (see ST1~ST11 in FIG. 10) has been computed the correlation matrix _{R p} from the MN set of the feature amount calculated by the attempt of the seating-Away the MN set For example, when the number of frames of 2 or more is set to f [frame] (f ≧ 2), the calculation is performed from the f-frame pressure distribution output during one set by trial of the MN set. It is also possible to calculate the correlation matrix R _p by using each feature amount of the total (MN × f) set. In this case, for example, when MN = 2 and f = 300, it is possible to calculate the correlation matrix R _p from 600 sets of feature values calculated by two sets of trials, as in the experimental example.

In (H016) the embodiment, in the authentication process (see ST101~ST112 in FIG. 11), one Mahalanobis _{D p} for each user from a single pressure distribution output by a set seating-Away of computed, but not limited to, for example, it is also possible to calculate the f-number of the Mahalanobis D _p for each user from the pressure distribution of f frames output in a set. In this case, for example, when f = 300, similarly to the experimental example, it is possible to calculate the 300 Mahalanobis D _p of which are computed by a set of trial, all 300 Mahalanobis D _p is the threshold value S when it becomes less as _D, to authenticate that the user of the p glance, when one or more Mahalanobis D _p is equal to or less than the threshold value S _D, or authenticated as a user of the p glance It is also possible.

C, NK ... personal authentication device, C2E, C5B ... feature amount calculation means, C2M1 ... correlation coefficient storage means, C5C ... distance calculation means, C5D ... threshold discrimination means, C5E ... registrant authentication means, C5H, C5H '... authentication display means, C7D ... correlation coefficient updating means, CH ... chairs, CH1 ... seat surface, D p _... Mahalanobis distance, FL ... floor, _{r ijp ...} correlation coefficient, S ... personal authentication system, S D _... threshold, SN ... pressure detection member, SN1 ... first detection unit, second detection unit, u _1p ... first feature amount, u _2p ... second feature amount.

Claims (6)

A chair on which the subject sits;
A first detector configured to detect a first pressure value, which is a pressure in the first detection region by the subject placed in a first detection region set in advance on the seat surface of the chair and seated; A second detector configured to detect a second pressure value, which is a pressure in the second detection region by the subject who is seated in the second detection region set in advance in a region shifted from the detection region. A pressure sensing member;
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
A personal authentication device comprising: When registering the registrant, the first pressure value and the second pressure value of the registrant detected by the pressure detection member, the average being 0 and the standard deviation being 1 The correlation coefficient calculated based on a reference space as an authentication standard of the registrant based on a set of combinations of a feature quantity and the second feature quantity having an average of 0 and a standard deviation of 1; The correlation coefficient storage means for storing the registrant in association with each other,
The personal authentication device according to claim 1, further comprising: When the registration of the registrant is updated, a combination of the first feature quantity and the second feature quantity calculated by the feature quantity calculation means when the registrant is authenticated by the registrant authentication means is included. Correlation coefficient updating means for associating the correlation coefficient recalculated based on the reference space with the registrant and storing the correlation coefficient in the correlation coefficient storage means;
The personal authentication device according to claim 2, further comprising: The floor on which the subject stands upright,
A first detector configured to detect a first pressure value, which is a pressure of a sole in the first detection area, arranged by the subject standing in an upright position in the first detection area set on the floor; A second detection unit that is arranged in a second detection region set in advance in a region deviated from one detection region and detects a second pressure value that is a pressure of a sole in the second detection region by the subject standing upright. And a pressure detection member having,
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
A personal authentication device comprising: A chair on which the subject sits;
A first detector configured to detect a first pressure value, which is a pressure in the first detection region by the subject placed in a first detection region set in advance on the seat surface of the chair and seated; A second detector configured to detect a second pressure value, which is a pressure in the second detection region by the subject who is seated in the second detection region set in advance in a region shifted from the detection region. A pressure sensing member;
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
A personal authentication system characterized by comprising: The floor on which the subject stands upright,
A first detector configured to detect a first pressure value, which is a pressure of a sole in the first detection area, arranged by the subject standing in an upright position in the first detection area set on the floor; A second detection unit that is arranged in a second detection region set in advance in a region deviated from one detection region and detects a second pressure value that is a pressure of a sole in the second detection region by the subject standing upright. And a pressure detection member having,
Based on the first pressure value and the second pressure value detected by the pressure detection member, a first preset value that specifies a relationship between the first pressure value and the second pressure value is specified. A first characteristic value that is a characteristic value of the first characteristic value, and a second characteristic that is set in advance to identify a relationship between the first pressure value and the second pressure value, wherein the first characteristic A feature amount calculating means for calculating a second feature amount that is a value of the different second feature;
Correlation coefficient storage means for storing in advance a registrant registered in advance and a correlation coefficient for each registrant that specifies the correlation between the first feature and the second feature;
The Mahalanobis distance of the subject is calculated based on the correlation coefficient stored in the correlation coefficient storage means and the calculated first feature quantity and second feature quantity of the subject. Distance calculating means to
Threshold determination means for determining whether or not the computed Mahalanobis distance of the subject is equal to or less than a preset threshold;
When it is determined that the Mahalanobis distance of the subject is equal to or less than the threshold, the subject is the registrant associated with the correlation coefficient used in the calculation of the Mahalanobis distance. Registrant authentication means to authenticate,
Authentication display means for displaying that the subject is the authenticated registrant;
A personal authentication system characterized by comprising: