Authentication Device And Authentication Method

• US 20100278397A1
• Filed: 08/27/2008
• Published: 11/04/2010
• Est. Priority Date: 08/31/2007
• Status: Active Grant

• Current Assignee: Fit Design System Company Limited
• Original Assignee: Fit Design System Company Limited
• Inventors: Tobe, Takeo, Kikuchi, Isao, Tsuji, Tetsuji, Watakai, Tadahito, Mizuno, Kenzo, Watanabe, Isao
• Granted Patent: US 8,194,942 B2
• Time in Patent Office: Days
• US Class Current: 382/124
• CPC Class Codes: A61B 5/1171 based on the shapes or appearances of their bodies or parts thereof
• CPC Class Codes: A61B 5/1172 using fingerprinting
• CPC Class Codes: A61B 5/489 Blood vessels
• CPC Class Codes: A61B 5/6826 Finger
• CPC Class Codes: A61B 5/6838 Clamps or clips
• CPC Class Codes: G06V 10/141 Control of illumination
• CPC Class Codes: G06V 40/1312 direct reading, e.g. contactless acquisition
• CPC Class Codes: G06V 40/1341 Sensing with light passing through the finger
• CPC Class Codes: G06V 40/14 Vascular patterns

Abstract

An authentication can be carried out with high accuracy even when there is a change in an amount of incident light, the air temperature, the body temperature, or a manner to place a finger on a scanner.

An authentication apparatus includes: a light source that emits an infrared light beam onto a finger; image pick-up means 8 that picks up an image of the finger by using a transmitted light beam, and thus outputs an image signal; low-frequency brightness component extracting means 61 that extracts a low-frequency brightness component signal from the image signal; middle-frequency brightness component extracting means 62 that extracts a middle-frequency brightness component signal from the image signal; correction means 63 that corrects the middle-frequency brightness component signal using the low-frequency brightness component signal; quaternary-value converting means 64 that converts the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a predetermined ratio; storage means 7 that stores a quaternary-valued image obtained by the quaternary value conversion; correlation coefficient calculating means 65 that calculates a correlation coefficient between a quaternary-valued evaluation image as an object of evaluation obtained by the quaternary value conversion, and a quaternary-valued registered image beforehand stored in the storage means, by using a pattern matching method based on a normalized correlation method; and judgment means 66 that judges whether or not the quaternary-valued evaluation image and the quaternary-valued registered image are images picked up from the same finger on a basis of the correlation coefficient.

Claims

1. An authentication apparatus comprising:

a light source that emits an infrared light beam onto a finger;

image pick-up means that picks up an image of the finger by using a transmitted light beam, and thus outputs an image signal;

low-frequency brightness component extracting means that extracts a low-frequency brightness component signal from the image signal by using a low-pass filter;

middle-frequency brightness component extracting means that extracts a middle-frequency brightness component signal from the image signal by using a middle-pass filter;

correction means that corrects the middle-frequency brightness component signal by using the low-frequency brightness component signal;

quaternary-value converting means that converts the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a predetermined ratio;

storage means that stores a quaternary-valued image obtained from the quaternary-value conversion by the quaternary-value converting means;

correlation coefficient calculating means that calculates a correlation coefficient between a quaternary-valued evaluation image as an object of evaluation and a quaternary-valued registered image by using a pattern matching method based on a normalized correlation method, the quaternary-valued evaluation image obtained from the quaternary-value conversion by the quaternary-value converting means, the quaternary-valued registered image being beforehand stored in the storage means; and

judgment means that judges whether or not the quaternary-valued evaluation image and the quaternary-valued registered image are images picked up from the same finger on a basis of the correlation coefficient which is calculated by the correlation coefficient calculating means.

2. The authentication apparatus according to claim 1, wherein the correlation coefficient calculating means calculates the correlation coefficient by using the following formula 1,

correlation coefficients=F/G  

 

[Formula 1]

where;

F represents a value of cross-correlation between the evaluation image and the registered image, and is defined as F=Σ

(I×

T);

G represents a value of autocorrelation of the registered image, and is defined as G=Σ

(T×

T);

I denotes a brightness of the evaluation image; and

T denotes a brightness of the registered image.

3. The authentication apparatus according to claim 1,

further comprising quaternary-valued reduced image generating means that generates a quaternary-valued reduced image by convolving a plurality of cells included in the quaternary-valued image obtained from the quaternary-value conversion by the quaternary-value converting means, wherein:

the correlation coefficient calculating means calculates a primary correlation coefficient between a quaternary-valued reduced evaluation image generated from the quaternary-valued evaluation image, and a quaternary-valued reduced registered image beforehand stored in the storage means, by using the pattern matching method based on the normalized correlation method; and

only when the primary correlation coefficient is not less than a predetermined value, the correlation coefficient calculating means calculates the correlation coefficient between the quaternary-valued evaluation image associated with the quaternary-valued reduced evaluation image, and the quaternary-valued registered image associated with the quaternary-valued reduced registered image, by using the pattern matching method based on the normalized correlation method.

4. The authentication apparatus according to any one of claim 1, further comprising a high-frequency brightness component extracting means that extracts a high-frequency brightness component signal from the image signal by using a high-pass filter, wherein:

the correction means corrects the middle-frequency brightness component signal and the high-frequency brightness component signal by using the low-frequency brightness component signal;

the quaternary-value converting means converts the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a first predetermined ratio, and converts the corrected high-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a second predetermined ratio;

the correlation coefficient calculating means calculates (1) a middle-frequency correlation coefficient between a quaternary-valued middle-frequency evaluation image obtained from the middle-frequency brightness component signal, and a quaternary-valued middle-frequency registered image beforehand stored, and (2) a high-frequency correlation coefficient between a quaternary-valued high-frequency evaluation image obtained from the high-frequency brightness component signal, and a quaternary-valued high-frequency registered image beforehand stored; and

on the basis of the middle-frequency correlation coefficient and the high-frequency correlation coefficient, the judgment means judges whether or not the quaternary-valued middle-frequency evaluation image and the quaternary-valued middle-frequency registered image are images picked up from the same finger, and whether or not the quaternary-valued high-frequency evaluation image and the quaternary-valued high-frequency registered image are images picked up from the same finger.

6. The authentication apparatus according to claim 2,

further comprising quaternary-valued reduced image generating means that generates a quaternary-valued reduced image by convolving a plurality of cells included in the quaternary-valued image obtained from the quaternary-value conversion by the quaternary-value converting means, wherein:

the correlation coefficient calculating means calculates a primary correlation coefficient between a quaternary-valued reduced evaluation image generated from the quaternary-valued evaluation image, and a quaternary-valued reduced registered image beforehand stored in the storage means, by using the pattern matching method based on the normalized correlation method; and

only when the primary correlation coefficient is not less than a predetermined value, the correlation coefficient calculating means calculates the correlation coefficient between the quaternary-valued evaluation image associated with the quaternary-valued reduced evaluation image, and the quaternary-valued registered image associated with the quaternary-valued reduced registered image, by using the pattern matching method based on the normalized correlation method.

7. The authentication apparatus according to any one of claim 2, further comprising a high-frequency brightness component extracting means that extracts a high-frequency brightness component signal from the image signal by using a high-pass filter, wherein:

the correction means corrects the middle-frequency brightness component signal and the high-frequency brightness component signal by using the low-frequency brightness component signal;

the quaternary-value converting means converts the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a first predetermined ratio, and converts the corrected high-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a second predetermined ratio;

the correlation coefficient calculating means calculates (1) a middle-frequency correlation coefficient between a quaternary-valued middle-frequency evaluation image obtained from the middle-frequency brightness component signal, and a quaternary-valued middle-frequency registered image beforehand stored, and (2) a high-frequency correlation coefficient between a quaternary-valued high-frequency evaluation image obtained from the high-frequency brightness component signal, and a quaternary-valued high-frequency registered image beforehand stored; and

on the basis of the middle-frequency correlation coefficient and the high-frequency correlation coefficient, the judgment means judges whether or not the quaternary-valued middle-frequency evaluation image and the quaternary-valued middle-frequency registered image are images picked up from the same finger, and whether or not the quaternary-valued high-frequency evaluation image and the quaternary-valued high-frequency registered image are images picked up from the same finger.

8. The authentication apparatus according to any one of claim 3, further comprising a high-frequency brightness component extracting means that extracts a high-frequency brightness component signal from the image signal by using a high-pass filter, wherein:

the correction means corrects the middle-frequency brightness component signal and the high-frequency brightness component signal by using the low-frequency brightness component signal;

the quaternary-value converting means converts the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a first predetermined ratio, and converts the corrected high-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a second predetermined ratio;

the correlation coefficient calculating means calculates (1) a middle-frequency correlation coefficient between a quaternary-valued middle-frequency evaluation image obtained from the middle-frequency brightness component signal, and a quaternary-valued middle-frequency registered image beforehand stored, and (2) a high-frequency correlation coefficient between a quaternary-valued high-frequency evaluation image obtained from the high-frequency brightness component signal, and a quaternary-valued high-frequency registered image beforehand stored; and

on the basis of the middle-frequency correlation coefficient and the high-frequency correlation coefficient, the judgment means judges whether or not the quaternary-valued middle-frequency evaluation image and the quaternary-valued middle-frequency registered image are images picked up from the same finger, and whether or not the quaternary-valued high-frequency evaluation image and the quaternary-valued high-frequency registered image are images picked up from the same finger.

5. An authentication method comprising the steps of:

emitting an infrared light beam onto a finger;

picking up an image of the finger by using a transmitted light beam, and thus outputting an image signal;

extracting a low-frequency brightness component signal from the image signal by using a low-pass filter;

extracting a middle-frequency brightness component signal from the image signal by using a middle-pass filter;

correcting the middle-frequency brightness component signal by using the low-frequency brightness component signal;

converting the corrected middle-frequency brightness component signal into quaternary values such that the quaternary values are distributed at a predetermined ratio;

storing a quaternary-valued image obtained by from the quaternary value conversion;

calculating a correlation coefficient between a quaternary-valued evaluation image as an object of evaluation, and a quaternary-valued registered image by using a pattern matching method based on a normalized correlation method, the quaternary-valued evaluation image obtained by from the quaternary value conversion, the quaternary-valued registered image being beforehand stored; and

judging whether or not the quaternary-valued evaluation image and the quaternary-valued registered image are images picked up from the same finger on a basis of the correlation coefficient.