JPH02263274A - Individual confirming device - Google Patents

Abstract

PURPOSE:To attain the highly accurate collation of individuals without causing the deterioration of the individual discriminating rate due to the deviation of a phase relation and to improve the individual confirming reliability by taking out the inter-finger phase information and evaluating it. CONSTITUTION:An individual confirming device consists of the means 11, 12 and 13 which produce the characteristic information on each finger, a storage means 17 which stores the characteristic information to be collated, and a discriminating means 20 which judges the individual confirmation based on the collation results on fingers obtained by the collation means 15 and 20 as well as the phase difference information detected by the phase difference detection means 15 and 20. Therefore the phase information on each finger can be extracted independently of each other and the inter-finger phase difference information obtained based on the phase information as well as the collation result of each finger can be integrately judged. Thus it is possible to collate individuals with each other with high accuracy and without deteriorating the reduction of the individual discriminating rate. Then the individual confirming reliability is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention (Field of Industrial Application) The present invention relates to a personal authentication device that authenticates an individual using physical characteristics such as a handprint or a fingerprint.

(Prior Art) Recently, devices that use the distance between finger joints have been developed as devices for personal authentication.

This type of personal authentication device uses an addition signal (feature information) obtained by adding pixel densities of finger images in a direction perpendicular to the longitudinal direction of the finger. That is, the density of the image information of the finger obtained from the difference between contact and non-contact between the finger and the finger rest on which the finger is placed is added in a direction perpendicular to the longitudinal direction of the finger to generate an addition signal. This is compared with an addition signal for comparison (characteristic information for comparison) registered in the dictionary in advance, and the comparison is performed depending on whether the distances between the joints in each signal match.

Note that with this type of device, processing may be performed on a single finger, or processing may be performed on two or more fingers with an increased amount of information in order to further improve discrimination ability. is possible.

By the way, in the case of a device that performs personal authentication using multiple fingers, the sum signal in the direction perpendicular to the longitudinal direction of the fingers is obtained for the entire image of the plural fingers, and the sum signal is Verification is performed by comparison. When multiple fingers are used in this way, the sum signal includes information about the six fingers as well as information representing the phase relationship between each finger, so from that point of view, the signal is better than when using one finger. It can also be said that the individual discrimination rate is high.

However, in this method, the information on the six fingers and the phase information cannot be extracted separately, so the individual discrimination rate may decrease. This occurs because the positional relationship of the six finger joints shifts slightly depending on how the fingers are placed, such as when the fingers are tilted, and this causes a shift in the phase relationship of the fingers. If the phase relationship of the fingers deviates in this way, the position of the concavity of the added signal corresponding to the indirect position of the finger will change, so even if the finger is from the same person (principal), it will not always be consistent with good reproducibility. It becomes impossible to obtain the addition signal. therefore,
In the conventional method of directly comparing summed signals, if multiple fingers are used, there is a high risk that the user will be identified as someone else (or that someone else will be identified as the person).

In this way, the conventional method is too sensitive to the phase relationship between each finger, and has the disadvantage of lacking accuracy, such as mistaking the person for another person due to subtle differences in the way the fingers are placed when inputting finger images. there were.

(Problems to be Solved by the Invention) As described above, the sum signal in the direction perpendicular to the longitudinal direction of the fingers is obtained for the entire image of multiple fingers, and the obtained one sum signal is used to authenticate the individual. In the case of equipment that performs
Accurate individual identification was not possible because the phase relationship between each finger was slightly shifted depending on how the fingers were placed.

Therefore, by extracting and evaluating the phase information between each finger, this invention can perform verification with high accuracy without causing a decrease in the individual identification rate due to a shift in the phase relationship, thereby increasing the reliability of authentication. The purpose of the present invention is to provide a personal authentication device that can significantly improve security.

[Structure of the Invention] (Means for Solving the Problem) The personal authentication device of the present invention includes an input means for inputting image information of a plurality of fingers, and an input means for inputting image information of a plurality of fingers inputted by the input means. 9 separation means that separates the image information corresponding to the fingers into every 6 fingers, and the density of each of the image information of the 6 fingers separated by this separation means is added in the direction orthogonal to the longitudinal direction of the fingers to separate the image information of the 6 fingers. a generation means for generating feature information; a storage means for storing feature information to be matched corresponding to the feature information generated by the generation means;
a collation means for collating the feature information generated by the generation means and the feature information to be collated stored in the storage means for every six corresponding fingers; a phase difference detection means for determining a phase matching position between information and the feature information to be matched, and detecting a phase difference between these pieces of information based on this phase matching position; and a matching result of six fingers obtained by the matching means. and a determining means for determining personal authentication based on the phase difference information detected by the phase difference detecting means.

(Function) This invention makes it possible to independently extract phase information for six fingers by the above-described means, and integrates the matching results of six fingers and phase difference information between each finger based on the phase information. I try to make judgments based on this.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 2 shows the personal authentication device of the present invention.

In FIG. 2, reference numeral 11 indicates two fingers Fa of the person to be authenticated when inputting finger image information.

This is a finger rest on which the Fb is placed, and is made of a transparent prism or the like. Image information of the fingers Fa and Fb placed on the finger rest 11 is obtained by irradiating light from the light source 12 disposed below the finger rest 11 through the finger rest 11. Camera 13 as a reflected light image
It can be read with .

The image information of the fingers Fa and Fb inputted to the camera 13 is converted into electrical signals, and then A/D converted by the A/D converter 14 and digitized. This digitalized finger image is stored in the image memory 1 by the control unit 15.
6 is temporarily stored.

On the other hand, the IC card 17 stores an arbitrary finger (in this embodiment, finger
, Fb) are stored. This feature information is an addition signal obtained by adding the pixel densities of the finger images in a direction perpendicular to the longitudinal direction of the finger. The IC card 17 is configured such that the characteristic information is written (at the time of registration) or read and verified by a reading/writing device 18 controlled by the control unit 15. The writing device 18 constitutes a dictionary section 19 for registering characteristic information of each person to be authenticated.

The control unit 15 obtains characteristic information of the fingers Fa and Fb and phase difference information between the fingers Fa and Fb from the finger images stored in the image memory 16, respectively.

Then, these feature information and phase difference information are compared with the feature information to be verified corresponding to the six fingers registered in the dictionary section 19, and thereby the identity is verified. In this case, the control unit 15 separates the images of fingers Fa and Fb using information (y direction addition signal) obtained by adding the densities of the finger images for the two fingers Fa and Fb in the longitudinal direction, The above matching process is performed using the fingers Fa and Fb as independent feature information. That is, the densities of the finger images for multiple (two) input fingers are added in the longitudinal direction of the fingers, the separation point (coordinates) of the fingers is determined from the resulting y-direction addition signal, and the separation point (coordinates) of the fingers is calculated at that position. Separate multiple commands into one command each. Then, feature information is obtained for each of the six fingers, and matching processing is performed using feature information to be matched that corresponds to each of the six fingers. Further, the phase difference information is obtained based on the matching position of the finger of the feature information and the feature information to be matched.

The host machine 20 controls, for example, the opening and closing of a door and the sounding of an alarm buzzer in accordance with the verification results from the control section 15. Note that this host machine 20
The subject of control differs depending on the security system to which this personal authentication device is applied.

Next, with reference to FIG. 3, the processing in the control section 15 will be explained in detail. Here, the index finger and middle finger of the right hand are used as fingers Fa and Fb, respectively, and it is assumed that these are determined by the will of the person to be authenticated, and that in principle only the person himself/herself can know the contents.

In FIG. 3(,a), fa and fb are finger images, that is, read by the camera 13, and are read by the A/D converter 1.
4 are image patterns of the index finger Fa and the middle finger Fb, respectively, which were digitized by A.4. These finger images fa
, fb is, for example, 51 lines/mm with a resolution of 10 lines/mm.
It is assumed that the image is captured in an area of 2 pixels x 512 pixels.

First, image separation of the index finger Fa and middle finger Fb will be explained. When the images of the two fingers Fa and Fb placed on the finger rest 11 are captured into the image memory 16, the images of the fingers fa and fb are transferred in the longitudinal direction X of the fingers, as shown in FIG. 3(b). By adding the densities, the y-direction addition signal Yab is obtained. This y-direction addition signal Yab usually contains high frequency components and is difficult to process as is, so the high frequency components are removed by passing it through a low pass filter, for example.

In the y-direction addition signal Yab obtained in this way, two peaks appear corresponding to the index finger Fa and the middle finger Fb. Therefore, by searching for the valley between two peaks,
Y coordinate value (separation point) y that separates two fingers Fa and Fb
t is found. This can be obtained, for example, by finding the minimum value of the y-direction addition signal Yab after passing through a low-pass filter.

On the other hand, when obtaining characteristic information about the index finger Fa and the middle finger Fb, the density of each finger image fa and fb is added in the direction y perpendicular to the longitudinal direction of the finger according to the separation point yt obtained above. A summation signal in the X direction is obtained for each. In other words, the pixels in the y direction (
512), by calculating the addition signal in the X direction in the range from 0 to the separation point yt, the feature information AA(X) for the index finger Fa can be obtained as shown in FIG.
Furthermore, by determining the addition signal in the X direction in the range from separation point yt to 511, as shown in FIG.
Feature information As(x) for the middle finger Fb is generated.

These feature information AA (X), As (x) are 6
There is a steep valley (minimum value) at the position of the horizontal wrinkle corresponding to the joint of the finger, and this is the parameter for personal authentication (
individuality).

Next, the operation of the personal authentication device configured as described above will be explained. In this device, there are two main processes: "registration processing" and "verification processing".

FIG. 4 is a flowchart for explaining the "registration process". For example, assume that the index finger Fa and middle finger Fb of the right hand of the person to be authenticated are placed on the finger rest 11 in the registration mode. Then, the image information based on the difference between contact and non-contact between the fingers Fa and Fb and the finger rest 11 is converted into an electric signal by the camera 13, and further converted into an electric signal by the A/D.
After being A/D converted and digitized by the converter 14, they are temporarily stored in the image memory 16 as finger images fa and fb (step 5T1).

The density of each of the finger images fa and fb is added in the longitudinal direction X of the finger in the control unit 15.

As a result, the y-direction addition signal Yab is generated as shown in FIG. 3(b). The y-direction addition signal Yab is passed through a low-pass filter (not shown) and converted into a smooth signal.
The separation point yt that separates b is found (step 5T2
).

When the separation point yt is determined in this way, the control unit 15
, the densities of the respective finger images fa and fb are added in the direction y perpendicular to the longitudinal direction X of the finger. That is, according to the separation point yt, addition signals of pixel densities in the X direction for the finger images fa and fb are respectively generated. As a result, as shown in FIG. 3(c), the index finger Fa
As shown in FIG. 3(d), feature information AB for the middle finger Fb is generated.
(X) is generated (step 5T3).

The feature information as the X-direction addition signal for the index finger Fa and the feature information as the X-direction addition signal for the middle finger Fb thus obtained are each subjected to appropriate band-pass filter processing and converted into a form that is easy to perform later matching processing. is registered in the dictionary section 19 (step 5T).
4). That is, the above characteristic information is transmitted to the IC card 1 of the person to be authenticated via the reading/writing device 18 under the control of the control unit 15.
7 respectively.

The registration process is completed as described above, and the feature information for verification regarding the fingers Fa and Fb of the person to be authenticated is transferred to the dictionary unit 19.
It will be registered in .

FIG. 1 is a flowchart for explaining the "verification process." Here, we first input a finger image (step 5T).
II), separation of finger images (step 5T12), and feature information A^ (x) for the index finger Fa and feature information AB for the middle finger Fb from the separated finger images.
The generation of (x) (step 5T13) is obtained by processing similar to those in the above-mentioned registration.

In this way, when the feature information AA (X) and As (x) are generated as the X-direction addition signals for the two fingers Fa and Fb, the feature information for the object to be matched is read out from the dictionary section 19 and Feature information for comparison and feature information AA(X) generated from the finger image.

AB (X) are aligned and verified for each corresponding finger (step 5T14).

Here, the feature information of the index finger Fa among the two feature information read from the dictionary section 19 is set as AAd(i), and the feature information of the index finger Fa generated from the input finger image is set as AAd(i).
(i). Then, the feature information AAd(i) and information AA obtained by shifting the feature information AA(i) by m
The addition result obtained by adding the squared error with (i+m) over a certain range is defined as SA (m)4. This addition result SA
(m) is expressed when m≧0 and m<0. In this case, the addition result S^ (m) is A
This is a parameter indicating the negative prose between Ad(t) and AA (i + m), and indicates that the smaller the addition result SA (m) is, the higher the degree of matching is, and takes a small value if it is the same person. Alignment is performed by varying m within a certain range, and m at which the value of SA (m) is the smallest is called the index finger misalignment element MA. It is assumed that alignment is achieved at this MA, and at this time Let the value of one prose S^(MA) of the added signal of the index finger be the result of the matching. The same process is performed for the middle finger Fb, and the middle finger deviation element MB,
The degree of coincidence Se(Me) of the added signals is determined.

Next, in step STI 5, phase matching is performed. This is done by calculating the phase-prose PD expressed by the following equation.

F D = l Ms MA 1 If the phase relationship between the middle finger and index finger of the input finger images during registration and verification is exactly the same, the phase matching positions of the six fingers will be the same between registration and verification, so MB mh4A. , the value of phase-prose PD in the above equation becomes zero. In this way, the more the phase relationships match, the smaller the value of phase-prose PD becomes.
In the case of the person himself/herself, the PD value will be small.

Next, in step 5T16, SA (MA) and SB (MB) determine whether the person to be authenticated is the person himself or herself or another person.
, and FD are combined.

For example, a well-known method for making a judgment using a combination is the following method using a discriminant function.

DF-aXsA (MA) +bXSB (MB) +cXPD Here, a, b, and c are constants.

As shown in this equation, by linear combination of three parameters, DF
Calculate new parameters. Constants a, b in the formula
, c are selected using statistical methods so as to maximize the ability to distinguish between the person and the other person. As this statistical method, for example,
Literature (“Multivariate analysis method”, author, Okuno et al., Japan Society of Science and Technology, 1
971, pp. 259 to 321), there is a method using the Mahalanobis distance.

A threshold value TH for identifying the person/other person is determined in advance, and whether the person to be authenticated is the person himself/herself or another person is determined based on whether DF does not exceed this TH. That is, DF≦TH...Discrimination as the person in question DF>TH... Discrimination as a stranger.

In addition to the above-mentioned approach, there is also a method of determining whether a person is a person or another person based on each of the three parameters and logically combining the results. In other words,
For SA (MA), a threshold value THA for identifying the person/other person is determined in advance, and the person/other person is determined as shown in the following equation.

SA (MA) ≦ THA ... Judgment with the person S^ (M
A)≦THA...Judged as someone else, also, SB (MB
), threshold values THB, THP are similarly set for PD.
is determined in advance, and the person/other person is determined as shown in the following equation.

SR(MB)≦TH,... Person and judgment SB(Me)
≦THB ・・・ Judgment with others PD5THP ・・
・ Judgment with the person P D > T Hp ... Judgment with others These S
The judgment results of A (M^), SB (MB), and PD are logically combined to finally determine whether the person is the person/other person.

For example, if the system requires high security, SA
(M^), SB (Ms), and PD, there are various ways of combining such methods, such as determining the person to be authenticated as the person only when the person is determined to be the person himself/herself, and determining others as others.

After the above verification process is completed, the result of the identification of the person/other person is sent to the host machine 20, and the host machine 2
At 0, processing is performed according to the determination result, such as opening the door if the person is identified as the person in question, or sounding an alarm buzzer if the person is identified as someone else. These processes differ depending on the security system to which the device is applied.

As described above, in a device that performs personal authentication using multiple fingers, it is possible to independently extract the phase matching position between the feature information for each floor and the feature information to be matched, and the matching results for each floor and the above The phase difference information between each finger based on the phase-matching position is comprehensively determined.

In other words, although the phase difference information between each finger includes individuality, the phase relationship may be slightly shifted depending on how the fingers are placed, and this phase shift may cause the person to be mistakenly identified as another person, or may be mistaken for another person. Because the ability to distinguish between a person and another person was being degraded, such as misidentification as the person, the phase difference information between each finger can be independently extracted and evaluated from the comparison result of the feature information of each floor and the feature information to be matched. ing. This makes it possible to prevent deterioration in the ability to discriminate between a person and another person due to subtle phase fluctuations between fingers when inputting finger images. Therefore, it is possible to eliminate problems such as the person in question being rejected as a stranger, or conversely in misidentifying another person as the person in question, making it possible to significantly improve the accuracy of personal authentication and the reliability of the device. It is.

Furthermore, the combination of multiple fingers is arbitrary and can be freely selected by the person to be authenticated, so the security is high.

In addition, in the above embodiment, the explanation was given using an example of a device that authenticates an individual using two fingers, but the invention is not limited to this, and can also be applied to a device that authenticates an individual using three or more fingers. Applicable.

It goes without saying that various other modifications can be made without departing from the gist of the invention.

[Effects of the Invention] As detailed above, according to the present invention, by extracting and evaluating the phase information between each finger, it is possible to achieve high accuracy without causing a decrease in the individual discrimination rate due to a shift in the phase relationship. It is possible to provide a personal authentication device that can perform verification and thereby significantly improve the reliability of authentication.

[Brief explanation of drawings]

The drawings show one embodiment of the present invention, and FIG. 1 is a flowchart shown to explain the flow of processing at the time of verification;
Figure 2 is a block diagram schematically showing an example of a personal authentication device, Figure 3 is a diagram used to explain the operation using two fingers as an example, and Figure 4 explains the flow of processing during registration. FIG. DESCRIPTION OF SYMBOLS 11... Finger rest, 12... Light source, 13... Camera, 15... Control part, 19... Dictionary part, Fa...
Index finger, Fb...middle finger. Applicant's Representative Patent Attorney Takehiko Suzue

Claims (2)

[Claims] (1) An input means for inputting image information of a plurality of fingers; a separation means for separating image information corresponding to the plurality of fingers inputted by the input means for each finger; and separation by the separation means. generating means for generating feature information for each finger by adding the density of each image information of each finger in a direction orthogonal to the longitudinal direction of the finger; and a matching target corresponding to the feature information generated by the generating means. a storage means for storing characteristic information for use, and a collation means for collating the characteristic information generated by the generation means and the characteristic information to be verified stored in the storage means for each corresponding finger. , a phase difference detection means for determining a phase matching position between the feature information and the to-be-matched feature information for each finger from the matching result of the matching means, and detecting a phase difference between these pieces of information based on this phase matching position; and a determining means for determining personal authentication based on the verification result of each finger obtained by the verification means and the phase difference information detected by the phase difference detection means. Device. (2) The separating means separates the image information corresponding to a plurality of fingers at the minimum value position of the added signal obtained by adding the density in the longitudinal direction of the fingers ( 1) Personal authentication device described.