WO2006006533A1

WO2006006533A1 - Personal identification system, personal identification method, personal identification program, and recording medium

Info

Publication number: WO2006006533A1
Application number: PCT/JP2005/012676
Authority: WO
Inventors: Atsuyoshi Nakamura; Mineichi Kudo; Jun Toyama; Hidetoshi Nonaka
Original assignee: National University Corporation Hokkaido University
Priority date: 2004-07-13
Filing date: 2005-07-08
Publication date: 2006-01-19
Also published as: JPWO2006006533A1

Abstract

A personal identification system for identifying a person without causing the person to perceive it by using information collected at low cost. In the system, personal feature extracting sensors (20-1 to no-n) extracts personal features such as the height and weight from the person to be identified, environment information sensors (30-1 to 30-m) detect environment information including the temperature, weather, and time, personal identification means (11) performs personal identification on the basis of the personal features and the environment information by using a personal identification rule (13). Candidate screening means (12) screens the candidates to be identified by the personal identification means (11) on the basis of high-accuracy personal identification information such as the password and the fingerprint from a high-accuracy personal identification device (70), user move sensing information by tracking sensors (60-1 to 60-k) sucha as infrared sensor and a pressure sensor, and information of a personal schedule (80). Personal identification rule updating means (14) updates the personal identification rule (13) by feeding back the identification result.

Description

Specification

Personal authentication system, personal authentication method, personal authentication program, and recording medium

Technical field

TECHNICAL FIELD [0001] The present invention relates to a personal authentication method, and particularly to a personal authentication system, a personal authentication method, a personal authentication program, and a recording medium that authenticate an individual by integrating a plurality of pieces of information obtained at low cost. .

Background art

Conventionally, as a personal authentication method, a method using a face image, voice, fingerprint, iris or other biological information has been developed. For example, in the method described in Patent Document 1, an individual is authenticated using a plurality of pieces of biological information.

Patent Document 1: Japanese Patent Laid-Open No. 08-16788

Disclosure of the invention

Problems to be solved by the invention

[0003] However, in the prior art disclosed in Patent Document 1, the user must be aware of the authentication operation such as bringing his face close to the camera or placing his finger on the sensor. There is a problem that the operation is both cumbersome and also has a great psychological influence such as making the user feel uneasy!

[0004] Also, the conventional authentication method is vulnerable to environmental changes during authentication. For example, if the illumination, its brightness, angle, etc. change, the accuracy of face authentication will decrease drastically, and fingerprint authentication may not be possible if your finger is injured.

[0005] Furthermore, when providing a personalized service for each family member on a television or the like used at home, security is so strict.

[0006] Therefore, the present invention provides a personal authentication system, a personal authentication method, a personal authentication program, and a recording medium capable of executing personal authentication using information obtained at low cost without making the user aware of it. The purpose is to provide.

Means for solving the problem [0007] In order to solve the above problems, a personal authentication system according to the present invention includes a personal feature extraction sensor for extracting a personal feature to be authenticated, an environmental information sensor for detecting the personal environmental information, and the personal Personal authentication means for authenticating the individual based on the characteristics and the environmental information.

The invention's effect

[0008] According to the present invention, the personal feature extraction sensor authenticates the individual based on the personal feature extracted from the individual to be authenticated, and the environmental information sensor detects the environmental information detected from the individual. It is possible to perform personal authentication by integrating multiple personal characteristics and environmental information, which are information that can be obtained at low cost.

Brief Description of Drawings

FIG. 1 is a block diagram showing a configuration of a personal authentication device according to an embodiment of the present invention.

[Fig.2] A diagram showing an example of a personal identification rule configured by a Bayesian network

[Fig.3] Diagram showing a specific example of personal authentication using a personal identification rule configured by Bayesian network

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of a personal authentication system 100 according to an embodiment of the present invention.

In FIG. 1, the personal authentication system 100 includes a personal authentication device 10, a plurality of personal feature extraction sensors 20-1 to 20-n (n is a natural number), and a plurality of environmental information sensors 30-1 to 30. —M (m is a natural number), clock 40, Internet 50, multiple tracking sensors 60—1 to 60—k (k is a natural number), high-precision personal identification device 70, and personal schedule 80 The

The personal authentication device 10 includes a personal authentication unit 11, a candidate narrowing unit 12, a personal identification rule 13, and a personal identification rule update unit 14.

[0014] The personal authentication means 11 includes personal feature information from a plurality of personal feature extraction sensors 20-l to 20-n, a plurality of environmental information sensors 30-l to 30-m, a clock 40, and the Internet 50. And the personal identification according to the personal identification rule 13 shown in Figure 2 below. I testify.

[0015] Candidate narrowing means 12 includes sensor outputs from a plurality of tracking sensors 60-l-60-k and authentication information from high-precision personal authentication device 70 (for example, an ID card, a password, a face image). , Fingerprint, iris) and candidates to be authenticated based on personal schedule 80. The candidate narrowing means 12 may be omitted when the number of candidates to be authenticated is not necessarily required for the system, for example, when the candidate to be authenticated is a family or the like. However, for example, the authentication information from the high-accuracy personal authentication device 70 can be used, and if there are many candidates to be authenticated, it is effective because the candidates to be authenticated can be narrowed down.

The personal identification rule 13 is stored in advance in a database (not shown), and is used by the personal authentication means 11 for authentication.

[0017] Since the personal identification rule update means 14 has begun to use this device, it is possible to correctly authenticate the user power when the authentication accuracy of the personal authentication device 10 is poor or when personal authentication is difficult. The result of failure is fed back by voice or the like, and the personal identification rule 13 used by the personal authentication means 11 is updated. This personal identification rule updating means 14 may also be omitted if the personal identification rule 13 which is not necessarily required for this system can be designed with high initial authentication accuracy. However, if it is difficult to authenticate the individual, such as when individual characteristics change, the parameters of the personal identification rule 13 are changed so that the authentication accuracy is improved while authenticating according to the personal identification rule 13, and personal identification is performed. This is effective because it can improve the authentication accuracy of Rule 13.

[0018] The plurality of personal feature extraction sensors 20-1 to 20-n are used to collect the user's personal feature information such as the individual's position, height, weight, sitting height, face width, shoulder height, odor, and clothing color. Obtain and output to personal authentication means 11. Further, the plurality of environmental information sensors 30-1 to 30-m acquire environmental information such as temperature, humidity, atmospheric pressure, sunshine, solar radiation, wind, rain, snow, and the like, and output them to the personal authentication means 11. The clock 40 measures the time and outputs it to the personal authentication means 11. In addition, the Internet 50 acquires weather information and other Web page information and outputs them to the personal authentication means 11.

Next, the operation of the personal authentication system 100 configured as described above will be described. In addition, this It is assumed that the personal authentication system 100 is installed in a company office or library.

First, the user is authenticated by the high-precision authentication device 70 using an ID card, a password, a face image, a fingerprint, and an iris and enters the office or the hall.

[0021] As the tracking sensors 60-1 to 60-k, for example, multiple infrared sensors are embedded in the ceiling of the office or in the building, or multiple pressure sensors are embedded on the floor. The movement of the user after entering the building is sensed by these tracking sensors 61-l-60-k. In addition, as the tracking sensors 60-1 to 60-k, it is also possible to grasp personal movement information using GPS (Global Positioning System).

[0022] Then, when the user intends to receive a certain service in an office or a certain place in the hall, the personal authentication system 100 performs personal authentication by the following processing.

[0023] First, in the personal authentication device 10, the position, height, weight, sitting height, face width, shoulders of the individual are detected by a plurality of personal feature extraction sensors 20-1 to 20-n arranged near the location. The user's personal characteristic information such as height, smell, and clothing color is acquired and output to the personal authentication means 11. If the position, height, face width, shoulder height, etc. of an individual are to be detected, distance sensors provided on the ceiling, walls, pillars, etc. in offices and buildings can be used. If you want to detect an individual's weight, you can use a pressure sensor embedded in the floor of the office or the building. If an individual's odor is detected, an odor sensor can be used. A color sensor can be used to detect the color of clothes. In addition, the information obtained from the distance sensor and the color sensor can extract the power of one or more cameras.

[0024] On the other hand, in the personal authentication device 10, a plurality of environmental information sensors 30-1 to 30-m arranged near the location simultaneously with the extraction of the personal features by the personal feature extraction sensors 20-1 to 20-n. By acquiring environmental information such as temperature, humidity, atmospheric pressure, sunlight, solar radiation, wind, rain, snow, etc. in the office or in the hall where the environmental information sensor is located, it is output to the personal authentication means 11. If the temperature is detected, a temperature sensor can be used. A humidity sensor can be used to detect humidity. A barometric sensor can be used to detect barometric pressure. A wind sensor can be used to detect wind. If it detects rain, a rain sensor can be used. If it detects snow, a snow sensor can be used. Furthermore, the clock 40 can be used for detecting the day of the week and the time. Also, The Internet 50 can be used to detect weather information and other Web page information.

[0025] As described above, the personal authentication means 11 uses the personal feature extraction sensors 20-1 to 20-n to determine the position, height, weight, sitting height, face width, shoulder height, smell, Acquire personal characteristics such as color, and acquire environmental information such as temperature, humidity, atmospheric pressure, sunlight, solar radiation, wind, rain, and snow from environmental information sensors 30-1 to 30-m, clock 40 and Internet 50.

Here, the personal characteristics and environment information acquired by the personal authentication means 11 of the present embodiment for performing personal authentication are an infrared sensor, a pressure sensor, an odor sensor, a personal feature extraction sensor such as an image processing result, etc. 20 — It is low-cost information that can be obtained without making the user aware of it using l-20-n. Therefore, the user does not need to perform troublesome operations or operations such as bringing his face close to the camera or inputting an ID or password for authentication. In addition, since the high-precision personal information such as fingerprints and irises from the high-precision personal authentication device 70 is used by the candidate narrowing down means 12 to narrow down the user candidates, the personal authentication means 11 Do not use accurate information for personal authentication. In the present embodiment, it is not essential to detect all of the individual position, height, weight, sitting height, face width, shoulder height, odor, clothing color, etc. Of course, it is also possible to detect one or more pieces of information.

[0027] The candidate narrowing means 12 includes high-precision authentication information such as fingerprints and irises from the high-precision personal authentication device 70, tracking information from the tracking sensors 60-1 to 60-k, and a database. From the personal schedule 80 stored in the service (not shown), the user candidates who receive the service at that location are narrowed down. It should be noted that the candidate narrowing down means 12 is not necessarily required to use all of the tracking information from the tracking sensors 60-l to 60-k and the personal schedule 80. Of course, you can use it in combination with the high-precision authentication information from the high-precision personal authentication device 70!

[0028] Then, in the personal authentication means 11, the personal features obtained from the personal feature extraction sensors 20-l to 20-n and the environmental information sensor 30-1 from the candidates narrowed down by the candidate narrowing-down means 12. Apply the environmental information obtained from ~ 30-m to the personal identification rule 13 stored in the database (not shown) to identify the individual. At that time, personal authentication means 1 1 is the personal information obtained from the personal feature extraction sensors 20-l to 20-n, such as personal position, height, weight, sitting height, face width, shoulder height, odor, clothing color, and environmental information sensor 30 — 1-30—m It is not mandatory to use all environmental information such as temperature, humidity, weather information, day of the week, time of day, etc. obtained from the clock, clock 40 and Internet 50. It is advisable to use personal features and use any one or more environmental information to identify the individual.

[0029] Here, if the authentication accuracy of the personal authentication device 10 is poor or if it is difficult to authenticate the individual because the personal identification rule 13 is used and started, in order to confirm whether the authentication has been performed correctly. In addition, in the personal authentication device 10 of the present embodiment, confirmation is made to the user, for example, by voice or display. The user refers to the voice, display, etc. to determine whether the authentication of the personal authentication means 11 is correct, and feeds back the result by voice, key operation, etc., and inputs it to the personal identification rule update means 14 To do.

[0030] The personal identification rule update means 14 updates the personal identification rule 13 used by the personal authentication means 11 based on the authentication result fed back by voice or key operation. As a result, even if the authentication accuracy of the personal authentication device 10 is poor or it is difficult to identify an individual because it is starting to use, the personal identification rule 13 is updated according to the user who uses this device. , It will be possible to authenticate correctly.

Here, a specific example of the personal identification rule used by the personal authentication means 11 for personal authentication will be described.

FIG. 2 is a diagram illustrating an example of a personal identification rule configured by a Bayesian network.

In the present embodiment, the personal identification rule 13 used by the personal authentication means 11 for personal authentication is constituted by a Bayesian network. A Bayesian network is a form of representation of the joint probability distribution of multiple variables, and it can be used to calculate the probability distribution of unknown variable values.

[0034] In the example of Fig. 2, variable nodes 1, Y2, ···, Ym corresponding to the temperature, humidity, weather, time, etc., which are the environmental information from the environmental information sensors 30-1 to 30-m, respectively. It has a personal identification variable node Z as a child, and as a child of the personal identification variable node Z, from the personal feature extraction sensor 2 0-1 to 20-n, individual height, weight, face width, shoulder height, etc. Corresponding This is a Bayesian network with a structure having individual characteristic information variable nodes XI, X2,.

[0035] In order to express the joint probability distribution by this Bayesian network, the probability P (Y),, (Υ

), ..., Ρ (Υ) and conditional probability ρ (ζ I γ, γ, ..., γ), ρ (χ I ζ), ρ (χ I ζ),

2 m 1 2 m l 2

..., ρ (χ _η I ζ) is required.

[0036] These probabilities can also estimate the appearance frequency power. At the time of personal authentication, personal characteristics and environmental information (X, Χ, ···, Χ, Υ, Υ, ···,)) = (χ, χ)

1 2 η 1 2 m 1 2

, ..., χ, y, y, ..., y) to posterior probability distribution Ζ (Ζ | χ, χ, ..., χ, y, y, · η 1 2 m 1 2 η 1 2

·, Y) can be calculated by the following (Equation 1).

[0037] [Equation 1]

I…,)… (Formula

∑ _z (Z | j '.jn:

[0038] If the set of variable Z values for the candidates narrowed down by the candidate narrowing means 12 is S, then the Z value z that maximizes the posterior probability in S can be found by the following (Equation 2). Is possible.

[0039] [Equation 2] z = argmaxP (z | ₁ , χ,, ..., χ „, y ₂ , ·", ^)… (Formula 2)

[0040] The personal authentication means 11 recognizes the user corresponding to z thus obtained as the user who is going to receive the service.

[0041] Then, in the personal identification rule update means 14, the output when the obtained personal characteristics and environmental information are input to the personal identification rule 13 is close to the personal identification result obtained by feedback. Update identification rule 13.

At that time, the updating method by the personal identification rule updating means 14 differs depending on what personal identification rule 13 is used. For example, in the case of Fig. 2, the probabilities P (Y), Υ (Υ), ...

1 2

Ρ (Υ) and conditional probability Ρ (Ζ I Υ, Υ, ···, Υ), P (X I Z), P (X I Z), ···, P (X m 1 2 ml 2

The estimated value of I z) will be updated. Updates to estimates are obtained through feedback. The inference result data can be estimated by adding to the existing data.

[0043] Here, the probabilities P (Y), Ρ (Υ), ···, Ρ (Υ) are not used in (Equation 1) and (Equation 2). · · · · Necessary when Y is unknown due to sensor failure. For example, if the value of Y is unknown, use (Equation 3) and (Equation 4) instead of (Equation 1) and (Equation 2).

[0044] [Equation 3]

[0045] [Equation ⁴ ] ( ^x i \ ^z )… (Formula ⁴ )

Next, a specific example of personal authentication using a personal identification rule configured by a Bayesian network will be shown.

[0047] For example, it is assumed that the personal characteristics and the environmental information of the past three times of the personal characteristics and the environmental information when the user Β and the user Β have come to the place where the personal authentication device 10 is arranged are as shown in Table 1. . Table 2 shows the personal characteristics and environmental information of the users who came to the location. In this example, for simplification, only height and weight are used as personal characteristics, and only weather, day of the week, and time are used as environmental information. In addition, it is assumed that the user candidates receiving the service are narrowed down to user A or user B by the candidate narrowing means 12.

[table 1]

User Height Weight Day of the week Time

A 165-170 60-63 Sunny Moon 16:00-17:00

A 165-170 57-60 Rain Fire 17: 00— 18:00

A 165-170 57-60 Rain Water 17:00-18:00

B 165-170 57-60 Sunny Tuesday 16:00-17:00

B 170-175 57-60 Rain Fire 16:00-17:00

B 165-170 57-60 Rain Fire 16: 00-17: 00 [Table 2]

[0048] In this case, with regard to height and weight, the characteristics of user A and user B are similar, and it is difficult to narrow down user A or user B for height and body gravity. However, it can be estimated from the weather, day of the week, and time that are the environmental information that the user receiving the service is likely to be the user B. Because user A has a rainy day that only came between 17:00 and 18:00 on Tuesday, whereas V has never come between 16:00 and 17:00. User B comes twice between 16:00 and 17:00 on Tuesday on a rainy day!

[0049] This estimation is actually automatically performed based on the identification rule learned from the data in Table 1. Specifically, if the conditional probabilities of various data in Table 1 are obtained by Laplace estimation, a Bayesian network as shown in Fig. 3 is created. When a certain probability and a certain probability of B are obtained, they become 1Z4 and 3Z4, respectively, and therefore it can be estimated that the probability of being a user power is high.

Here, a specific example in which personal authentication is performed using only weather information, day of the week, and time as environment information has been shown, but the present invention is not limited to this. For example, the user power will come to the place where the personal authentication device 10 is installed or will not come, or the time required to come to the place where the personal authentication device 10 is placed will be greatly affected by the result of a sport game being broadcast on television. If it is possible to use it, it is possible to use the information on the game results that can be obtained from a specific page of the Internet as environmental information.

[0051] Although a specific example in the case where it is difficult to perform personal authentication from personal characteristics is shown here, of course, the present invention is not limited to this. For example, even if personal authentication can be performed using only the personal characteristics of height and weight, it is possible to correctly perform personal authentication using an identification rule that uses both environmental information and personal information. In other words, personal authentication is performed using information that is effective for personal authentication of the user from the acquired personal characteristics and environmental information.

As described above, in the personal authentication system 100 of the present embodiment, the personal feature extraction sensor 2 0— 1 to 20 — n personal characteristics such as height, weight, face width, shoulder height, etc. extracted from individuals to be authenticated, environmental information sensor 30—l-30—m Since individuals are authenticated based on environmental information such as weather, time, etc., individual authentication is performed by integrating multiple personal features and environmental information, which are information that can be obtained at low cost without making the user aware of it. Can be executed.

[0053] As a result, the user can receive a personalized service without performing a special operation for authentication, that is, without being aware of the authentication, thereby improving user convenience. Can do.

[0054] Further, in the personal authentication system 100 of the present embodiment, the candidate narrowing down means 12 uses high-precision personal authentication information such as an ID card, a personal identification number, a face image, and a fingerprint from the high-precision personal authentication device 70, and Tracking information such as infrared sensors embedded in the ceiling and pressure sensors embedded in the floor 60-l-60-k user movement detection information, and information stored in a database etc. Therefore, the personal authentication means 11 narrows down the individual candidates to be authenticated, so that the number of individual candidates to be authenticated by the personal authentication means 11 decreases, and the speed and burden of the authentication process is reduced. Can be reduced.

[0055] Further, in the personal authentication system 100 of the present embodiment, since the personal authentication means 11 inputs various personal characteristics and environmental information into the personal identification rule 13 by the Bayesian network, personal identification is performed. Personal authentication can be performed accurately and easily.

[0056] Furthermore, in the personal authentication system 100 of the present embodiment, the personal identification rule update means 14 feeds back the personal authentication result of the personal identification rule 13 and updates the personal identification rule 13. Even if the authentication accuracy of the personal authentication device 10 is low or it is difficult to identify an individual when using this device, it is possible to adaptively change the number of personal identification rules 13 through actual personal authentication processing. Therefore, the accuracy of personal authentication can be improved.

In the present embodiment, as shown in FIG. 1, the configuration of personal authentication system 100 has been described in terms of hardware by showing a block diagram, but the present invention is not limited to this. For example, a CPU and a node data that stores a program for executing the CPU as described above. Of course, a general-purpose computer-powered personal authentication system 100 such as a PC having a storage device such as a disk memory may be executed in software. In this case, the personal authentication program for performing the function as the personal authentication system 100 can be recorded on a recording medium such as a CD, a flash memory, or a USB memory. When using a personal authentication program, the recording medium strength personal authentication program is read and installed in a storage device in the computer, or downloaded and stored from a server or the like via a network such as the Internet. Store in the device. In this way, a general-purpose computer such as a PC can execute the personal authentication described above by executing the personal authentication program that functions as the personal authentication system 100 described above.

[0058] This specification is based on Japanese Patent Application No. 2004-206380 filed on Jul. 13, 2004. All this content is included here.

Industrial applicability

[0059] The personal authentication system, personal authentication method, personal authentication program, and recording medium according to the present invention can execute personal authentication using information obtained at low cost without making the user aware of it. It is useful as a personal authentication system, a personal authentication method, a personal authentication program, and a recording medium for authenticating an individual by integrating a plurality of information. It is also useful for providing personalized services to users of home appliances connected to multiple home appliances such as TVs, personal computers, refrigerators, lighting, and air conditioning.

Claims

The scope of the claims

[1] A personal feature extraction sensor that extracts the personal features to be authenticated;

An environmental information sensor for detecting the personal environmental information;

A personal authentication system comprising: personal authentication means for authenticating the individual based on the personal characteristics and the environmental information.

[2] The apparatus further comprises candidate narrowing means for narrowing down candidates to be authenticated by the personal authentication means based on at least one of the high-precision personal authentication information, movement information and schedule information of the individual. Item 1. Personal authentication system.

[3] The personal authentication means performs personal authentication according to a personal identification rule expressed by a Bayesian network having a personal identification variable node as a child of the environment information variable node and a personal characteristic variable node as a child of the personal identification variable node. The personal authentication system according to claim 1 to be performed.

4. The personal authentication system according to claim 3, further comprising a personal identification rule updating unit that feeds back a personal authentication result based on the personal identification rule and updates the personal identification rule.

[5] The personal feature extraction sensor includes a sensor that detects the position of the individual, a sensor that detects the height of the individual, a sensor that detects the weight of the individual, a sensor that detects the sitting height of the individual, and a face of the individual Including at least one of a sensor for detecting a width, a sensor for detecting a height of a shoulder of the individual, a sensor for detecting the smell of the individual, and a sensor for detecting a color of the clothes of the individual,

The environmental information extraction sensor is a sensor that detects time, a sensor that detects air temperature, a sensor that detects humidity, a sensor that detects atmospheric pressure, a sensor that detects sunlight, a sensor that detects sunlight, a sensor that detects wind, The personal authentication system according to claim 1, comprising at least one of a sensor that detects rain, a sensor that detects snow, and a sensor that detects information obtained from the Internet.

[6] extracting individual characteristics to be authenticated;

Detecting the personal environmental information;

Authenticating the individual based on the personal characteristics and the environmental information.

[7] extracting individual characteristics to be authenticated;

Detecting the personal environmental information;

A step of authenticating the individual based on the personal characteristics and the environmental information;

[8] extracting individual characteristics to be authenticated;

Detecting the personal environmental information;

A step of authenticating the individual based on the personal characteristics and the environmental information, and a recording medium on which a personal authentication program is executed for causing a computer to execute.