JP2004167227A - Personal identification method, iris registration device, iris identification device, and personal identification program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To secure sufficiently high identification precision in a personal identification using iris identification even under various conditions under outdoor light, in a dark place, or the like. <P>SOLUTION: The personal identification method is executed by registering a plurality of iris codes together with a ratio R of pupil diameter/iris diameter for each registered person in an iris database . Upon identification, features are extracted from a photographed iris image to obtain an iris code and calculate the ratio R of pupil diameter/iris diameter. The ratio R in the registration side and the calculated ratio R are compared for identification and a matched iris code is selected for the identification as the reference object from the iris database. <P>COPYRIGHT: (C)2004,JPO

Description

  The present invention relates to a technique for personal authentication using iris recognition, and particularly to a technique for improving the accuracy of iris authentication in various situations such as under external light such as sunlight or at night.

  In recent years, the technology of personal authentication using iris recognition has begun to be used for entry / exit management to important facilities, ATM (Automated Teller Machine) of banks and the like, PC log-in applications, and the like.

  Patent Literature 1 discloses a basic iris recognition method in which iris image information obtained at the time of authentication is compared with stored iris image information to identify an individual. In this method, the pupil diameter is changed by illuminating the eyes, the illumination intensity is controlled to make the pupil a predetermined diameter, and then the iris images or feature amounts extracted from the iris images are compared. . The diameter of the pupil changes depending on the surrounding brightness, emotion, and the like. However, if the pupil diameters are different from each other, it becomes difficult to compare iris images or feature amounts. For this reason, in Patent Literature 1, the pupil diameter is adjusted by illumination and the comparison is made. However, it takes a certain amount of time to change the pupil diameter to a predetermined diameter, and therefore, there is a problem that authentication takes time.

  On the other hand, Patent Documents 2 and 3 solve the above problem as follows.

  Patent Document 2 proposes a feature extraction method that does not depend on the pupil diameter by expressing the iris region in a polar coordinate system. That is, the center of the pupil / iris boundary approximated by a circle is used as the origin of the polar coordinates, and the radial coordinates are expressed as the ratio of the distance between the pupil / iris boundary and the iris / sclera boundary. As a result, even when the pupil diameter is different between the registration and the authentication, a feature of a constant expression is extracted, and the feature amounts can be compared. Therefore, it is not necessary to control the pupil diameter at the time of authentication, and the authentication time can be reduced as compared with Patent Document 1.

In Patent Document 3, at the time of registration, a plurality of iris images having different pupil diameters are photographed by changing the illumination intensity, and feature amounts extracted from the plurality of iris images are registered, and the iris photographed at the time of authentication is registered. The feature amount extracted from the image is compared with a plurality of registered feature amounts. Since a plurality of iris images having different pupil diameters are photographed at the time of registration, even if the pupil diameter of the iris image at the time of authentication is an arbitrary value, comparison is made with registered data having substantially the same pupil diameter. be able to. Therefore, it is not necessary to control the pupil diameter at the time of authentication, and the authentication time can be reduced as compared with Patent Document 1.
Japanese Patent Publication No. 5-84166 Japanese Patent Publication No. Hei 8-504979 JP 2000-194855 A

  Now, Patent Document 2 mentioned above states that "the iris elongates and contracts in response to light to adjust the size of the pupil, but its fine tissue hardly changes except for the elongation and contraction." Have been. By the polar coordinate expression, even if the pupil diameter changes, the iris pattern hardly changes, so that authentication with a reduced rejection rate is possible. However, strictly speaking, the iris pattern changes slightly, and this change causes a slight increase in the Hamming distance at the time of authentication. Even if the increase in the hamming distance is within an allowable range, if the hamming distance is combined with other factors that increase the hamming distance, such as hiding of the iris area by eyelids and eyelashes, the person may sometimes be rejected.

  In Patent Literature 3, since the feature data extracted from a plurality of iris images is registered at the time of registration, the capacity of the registration database increases. In addition, at the time of authentication, since it is necessary to compare with the plurality of feature data, the processing time at the time of authentication increases.

  In view of the above-described problem, the present invention provides a method of performing personal authentication using iris authentication, which is robust against fluctuations in pupil diameter and suppresses an increase in the number of databases and an increase in processing time during authentication. Make it an issue.

  In order to solve the above-described problem, the present invention provides, as a personal authentication method using an iris image, at the time of registration, an iris image is acquired for a registered person, and feature data and a pupil opening index are obtained from the acquired iris image. Using the obtained feature data and the determined pupil opening index, data registration for the registrant is performed in the iris database.At the time of authentication, an iris image is obtained for the certifier, and feature data is obtained from the obtained iris image. And the pupil opening index, refer to the data registered for the registrant in the iris database, obtain the matching target feature data using the pupil opening index obtained at the time of authentication, the matching target feature data By comparing the authentication data with the characteristic data obtained at the time of authentication, it is determined whether or not the user is the same as the registrant.

  According to the present invention, at the time of registration, data registration using the feature data obtained from the iris image and the pupil opening is performed in the iris database. Then, at the time of authentication, the data registered in the iris database is referred to, and the characteristic data to be collated is obtained using the pupil opening degree index obtained from the iris image at the time of authentication. For this reason, even if the pupil at the time of authentication is in a miotic state or a mydriatic state, feature data corresponding to the degree of opening of the pupil can be obtained as a collation target. Sufficiently accurate personal authentication with a reduced rejection rate can be executed.

  Then, in the personal authentication method according to the present invention, at the time of registration, the feature data is registered in the iris database in association with the registrant together with the pupil opening index, and is registered in the iris database at the time of authentication. By comparing the pupil opening index registered along with the feature data and the pupil opening index obtained at the time of authentication from the feature data registered in association with the Preferably, it is specified.

  Thereby, at the time of registration, the feature data obtained from the iris image is registered in the iris database together with the pupil opening index. Then, at the time of authentication, feature data to be collated is specified by comparing pupil opening indices among feature data registered in the iris database. Therefore, irrespective of the state of the pupil at the time of authentication, feature data corresponding to the degree of opening of the pupil is specified as a collation target. In addition, by using the pupil opening index, it is possible to easily search for feature data to be collated, so that authentication can be performed in a short time.

  Further, in the personal authentication method according to the present invention, at the time of registration, a plurality of iris images having different pupil opening degrees for the registered person are obtained, and a plurality of feature data respectively extracted from the obtained plurality of iris images and A relational expression between the feature data and the pupil opening degree index is obtained from the pupil opening degree index, and a parameter for expressing the relational expression is registered in the iris database in association with the registrant, and at the time of authentication, the iris database It is preferable to obtain a relational expression from parameters registered in association with the registrant, and substitute the pupil opening index obtained at the time of authentication into this relational expression to obtain characteristic data to be collated.

  Thereby, at the time of registration, a parameter for expressing a relational expression between a plurality of feature data obtained from a plurality of iris images having different pupil opening degrees and a pupil opening index is registered in the iris database. Then, at the time of authentication, the pupil opening index obtained from the iris image at the time of authentication is substituted into the relational expression represented by the parameters registered in the iris database, and the matching target feature data is obtained. Therefore, irrespective of the state of the pupil at the time of authentication, feature data corresponding to the degree of opening of the pupil is obtained as a matching target. In addition, since the characteristic data to be collated can be easily obtained only by substituting the pupil opening index into the relational expression, the authentication can be performed in a short time.

  In the personal authentication method according to the present invention, at the time of registration, a plurality of iris images having different pupil opening degrees for the registered person are obtained, and a plurality of feature data obtained from the obtained plurality of iris images is obtained. Determine the feature data to be registered, determine a conversion rule for performing conversion from the registered feature data to other feature data having a different pupil opening index, and determine the registered feature data and the conversion rule with the registrant. Registered in association with the iris database, and at the time of authentication, based on feature data and conversion rules registered in association with the registrant in the iris database, collation is performed using the pupil opening index obtained at the time of authentication. Preferably, the target feature data is generated.

  By this, at the time of registration, a conversion rule from a plurality of feature data obtained from a plurality of iris images having different pupil opening degrees to registered feature data and other feature data having a different pupil opening index is obtained, and the iris is obtained. Registered in the database. Then, at the time of authentication, based on the feature data registered in the iris database and the conversion rule, feature data to be collated is generated using the pupil opening degree index obtained from the iris image at the time of authentication. Therefore, irrespective of the state of the pupil at the time of authentication, feature data corresponding to the degree of opening of the pupil is generated as a matching target.

  In addition, the present invention provides a personal authentication method using an iris image, a first step of obtaining an iris image of a person to be authenticated, and feature data and a pupil opening index from the iris image obtained in the first step. A second step of obtaining the pupil opening index, and referring to data registered for the registrant in the iris database in which data registration using the pupil opening index is performed, and using the pupil opening index obtained in the second step. A third step of obtaining feature data to be matched, and comparing the feature data of the matching object obtained in the third step with the feature data obtained in the second step, whereby the person to be authenticated is And a fourth step of determining whether the user is the same as the registrant.

  The present invention also provides a device for performing data registration for iris authentication, a device for obtaining an iris image of a registered person, a device for obtaining feature data and a pupil opening index from the iris image, Means for registering data on the registrant in the iris database using the pupil opening index.

  Further, the present invention, as an apparatus for performing personal authentication using an iris image, means for acquiring an iris image for a person to be authenticated, means for obtaining feature data and a pupil opening index from the iris image, Means for obtaining feature data to be checked by referring to data registered in association with a registrant in an iris database in which data registration using the pupil opening index is performed using the pupil opening index; Means for comparing the characteristic data with the characteristic data to determine whether the person to be authenticated is the same as the registrant.

  Further, the present invention, as a program for causing a computer to execute personal authentication using an iris image, a step of obtaining feature data and a pupil opening index from an iris image obtained for a subject, and using the pupil opening index Referring to data registered for a registrant in the iris database in which data registration using the pupil opening index has been performed, and obtaining feature data to be matched; and the feature data to be matched and the feature data. And determining whether the person to be authenticated is the same as the registrant by comparing the registrant with the computer.

  According to the present invention, even if the pupil at the time of authentication is in a miotic state or a mydriatic state, feature data corresponding to the degree of opening of the pupil can be obtained as a matching target, so even in various situations. Therefore, it is possible to execute sufficiently accurate personal authentication. Further, by using the pupil opening index, it is possible to easily search or acquire the feature data to be collated, so that the authentication can be performed in a short time.

  In the first aspect of the present invention, as a personal authentication method using an iris image, at the time of registration, an iris image is obtained for a registered person, and characteristic data and a pupil opening index are obtained and obtained from the obtained iris image. Using the feature data and the pupil opening index, register the data of the person to be registered in the iris database, obtain an iris image of the person to be authenticated at the time of authentication, and obtain feature data and pupil opening from the obtained iris image. An index is obtained, the data registered for the registrant in the iris database is referred to, the characteristic data to be collated is obtained using the pupil opening index obtained at the time of authentication, and the characteristic data of the collation object and the data at the time of authentication are obtained. A method is provided for determining whether or not the person to be authenticated is the same as the registrant by comparing the obtained characteristic data with the registrant.

  In the second aspect of the present invention, at the time of registration, the feature data is registered in the iris database in association with the registrant together with the pupil opening index, and at the time of authentication, the registrant is associated with the iris database. By comparing the pupil opening index registered together with the feature data with the pupil opening index obtained at the time of authentication, the characteristic data to be matched is specified from the registered feature data. A personal authentication method according to a first aspect is provided.

  In the third aspect of the present invention, at the time of registration, at least three pieces of feature data obtained from the iris images in the miotic state, the normal state, and the mydriatic state are registered for the registrant at least. Provide personal authentication method.

  According to a fourth aspect of the present invention, at the time of registration, a plurality of iris images having different pupil opening degrees are obtained for the registered person, and feature data is respectively obtained from the obtained plurality of iris images. A personal authentication method according to a second aspect of selecting feature data to be registered in the iris database from the plurality of feature data by matching each other is provided.

  According to a fifth aspect of the present invention, in the second aspect, at the time of authentication, the authentication process is stopped when feature data having a pupil opening index that is close to the obtained pupil opening index by a predetermined degree is not registered for the registrant. Provide personal authentication method.

  According to the sixth aspect of the present invention, when the authentication process is stopped, a preferable iris image photographing condition is estimated based on the pupil opening index obtained at the time of authentication and the pupil opening index related to the registered feature data. A fifth aspect of the present invention provides a personal authentication method that prompts the subject to reacquire an iris image under the estimated photographing conditions.

  In the seventh aspect of the present invention, at the time of registration, a plurality of iris images having different pupil opening degrees for the registered person are obtained, and a plurality of feature data and a pupil opening index respectively obtained from the obtained plurality of iris images. A relational expression between the feature data and the pupil opening degree index is obtained from, and a parameter for expressing this relational expression is registered in the iris database in association with the registrant, and at the time of authentication, a registrant is registered in the iris database. A personal authentication method according to a first aspect is provided in which a relational expression is obtained from parameters registered in association with each other, and a pupil opening degree index obtained at the time of authentication is substituted into the relational expression to obtain the characteristic data to be collated.

  According to an eighth aspect of the present invention, there is provided the personal authentication method according to the seventh aspect, in which, at the time of registration, the parameters are thinned and registered, and at the time of authentication, the thinned parameters are restored by interpolation.

  According to a ninth aspect of the present invention, at the time of registration, a plurality of iris images having different pupil opening degrees for the registered person are acquired, and registered based on a plurality of feature data obtained from the acquired plurality of iris images. Along with specifying the data, a conversion rule for performing a conversion from the registered feature data to another feature data having a different pupil opening index is obtained, and the registered feature data and the conversion rule are associated with the registrant and the Registered in the iris database, at the time of authentication, based on the feature data and conversion rules registered in association with the registrant in the iris database, using the pupil opening index obtained at the time of authentication, the feature data of the matching target Is provided.

  According to a tenth aspect of the present invention, there is provided the personal identification method according to the first aspect, wherein the pupil opening index is a ratio between a pupil diameter and an iris diameter in an iris image.

  According to an eleventh aspect of the present invention, as a personal authentication method using an iris image, a first step of acquiring an iris image of a person to be authenticated, and feature data and a pupil opening degree from the iris image acquired in the first step. A second step of obtaining an index, and referring to data registered for a registrant in an iris database in which data registration using the pupil opening index has been performed, and calculating the pupil opening index obtained in the second step. A third step of obtaining feature data to be collated using the method, and comparing the feature data of the collation object obtained in the third step with the feature data obtained in the second step to obtain the person to be authenticated. And a fourth step of determining whether or not the registrant is the same as the registrant.

  In a twelfth aspect of the present invention, in the iris database, feature data of an iris image for each registrant is registered together with at least one pupil opening index, and in the third step, the iris database The pupil opening index registered together with the feature data and the pupil opening index obtained in the second step are compared from at least one feature data registered in association with the registrant. Thereby, the personal authentication method according to the eleventh aspect for specifying the characteristic data to be collated is provided.

  In a thirteenth aspect of the present invention, in the iris database, for each registrant, a parameter representing a relational expression between the feature data of the iris image and the pupil opening index is registered, and in the third step, Obtaining a relational expression from the parameters registered in the iris database in association with the registrant, and substituting the pupil opening index obtained in the second step into the relational expression obtains the characteristic data of the matching target. The personal authentication method according to the eleventh aspect is provided.

  In a fourteenth aspect of the present invention, the iris database includes, for each registrant, feature data of an iris image and a conversion rule for performing conversion from the feature data to other feature data having a different pupil opening index. The pupil opening index obtained in the second step based on the feature data and the conversion rule registered in the third step in association with the registrant in the third step. An eleventh aspect of the present invention provides a personal authentication method for generating feature data to be compared using the personal authentication method.

  According to a fifteenth aspect of the present invention, as an apparatus for performing data registration for iris authentication, means for acquiring an iris image of a registered person, means for obtaining feature data and a pupil opening index from the iris image, Means for registering data on the registrant in the iris database using the feature data and the pupil opening index.

  According to a sixteenth aspect of the present invention, as an apparatus for performing personal authentication using an iris image, means for acquiring an iris image of a subject, and means for obtaining feature data and a pupil opening index from the iris image Using the pupil opening index, referring to data registered in association with a registrant in an iris database in which data registration using the pupil opening index has been performed, and a means for obtaining feature data to be matched; Means for comparing the characteristic data to be collated with the characteristic data to determine whether or not the person to be authenticated is the same as the registrant.

  In a seventeenth aspect of the present invention, as a program for causing a computer to execute personal authentication using an iris image, a step of obtaining feature data and a pupil opening index from an iris image obtained for a subject, With reference to the data registered for the registrant in the iris database that has been registered data using the pupil opening index, to obtain the characteristic data of the collation target, the characteristic data of the collation target, Determining whether the authenticated person is the same as the registrant by comparing the characteristic data with the registrant.

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

(1st Embodiment)
FIG. 1 is a diagram conceptually showing a personal authentication method according to the first embodiment of the present invention. As shown in FIG. 1, in the present embodiment, a plurality of iris codes and a ratio R between pupil diameter and iris diameter (hereinafter referred to as “pupil diameter / iris diameter ratio”) are written in the iris database 12 for each registrant. ). Then, at the time of authentication, an iris code is obtained by feature extraction from the captured iris image, and a pupil diameter-iris diameter ratio R is obtained. Then, by comparing the ratio R on the registration side with the ratio R at the time of authentication, an appropriate iris code is selected as a collation target, and authentication is performed.

  FIG. 2 is a diagram illustrating the overall configuration of the authentication system according to the present embodiment. In FIG. 2, an iris authentication server 11 has an iris database 12 for storing feature data of iris images of a plurality of registrants, and is connected to a network 13 such as the Internet, a dedicated line, or a public line. In addition, the iris registration device 14 and the iris authentication device 15 are similarly connected to the network 13. The iris registration device 14 transmits the iris code generated at the time of registration to the iris database 12. The iris authentication device 15 performs personal authentication by comparing the iris code generated at the time of authentication with the iris code at the time of registration acquired from the iris database 12.

  Note that a plurality of iris authentication servers 11 may be provided for each region or institution used, or a plurality of servers including a mirror server for distributing the load. The iris database 12 may be connected to the iris authentication server 11 via a network.

  FIG. 3 is a block diagram showing a configuration of the iris registration device and the iris photographing device in the authentication system of FIG. As shown in FIG. 3, the iris photographing device 16 includes a lighting unit 31 that performs visible light illumination around the face of the registrant, a photographing unit 32 that can continuously photograph a plurality of iris images of the registrant, and An illumination control unit 33 that controls the emission intensity of the illumination 31 is provided.

  In addition, the iris registration device 14 includes an iris image acquisition unit 34 for acquiring an iris image, a feature extraction unit 35 for extracting feature data (iris code) from the acquired iris image, and a pupil opening index in the acquired iris image. A ratio calculation unit 36 for calculating the pupil diameter / iris diameter ratio, and a characteristic transmission unit 37 for transmitting the characteristic data and the pupil diameter / iris diameter ratio to the network 13 in association with each other. Here, the iris image acquisition unit 34 acquires a plurality of iris images transmitted from the iris imaging device 16 with different pupil opening degrees. The feature data transmitted from the feature transmitting unit 37 to the network 13 is registered in the iris database 12.

  When the iris registration device 14 is configured integrally with the iris authentication server, the feature data is registered in the iris database in association with the feature data and the pupil diameter / iris diameter ratio instead of the feature transmission unit 37. What is necessary is just to provide a storage part. Further, the iris registration device 14 and the iris photographing device 16 may be configured integrally or separately. Further, the iris registration device 14 may perform only photographing of the iris image, transmit the photographed iris image to the iris authentication server 11, and generate the iris code on the iris authentication server 11 side. Further, all functions of the iris registration device 14 may be built in the personal authentication server 11.

  FIG. 4 is a diagram showing the configuration of the iris authentication device in the authentication system of FIG. As shown in FIG. 4, the iris authentication device 15 includes an iris image acquisition unit 41 that acquires an iris image of a person to be authenticated, a feature extraction unit 42 that extracts feature data from the acquired iris image, and an acquired iris image. The ratio calculating unit 43 for calculating the pupil diameter / iris diameter ratio in the iris database 12 refers to the iris database 12 via the network 13 and compares the pupil diameter / iris diameter ratio to select and acquire the characteristic data to be matched. A quantity acquisition unit 44 and an authentication unit 45 that performs authentication by comparing feature data to be collated with acquired feature data are provided.

  In addition, the iris authentication device 15 may perform only the photographing of the iris image for authentication, transmit the photographed iris image to the iris authentication server 11, and generate the iris code in the iris authentication server 11. Further, the comparison with the registered iris code may be performed in the iris server 11, and the iris authentication device 15 may receive only the authentication result. The iris authentication device 15 has all the functions of the iris authentication server 11, the iris database 12, the iris registration device 14, and the iris photographing device 16 in FIG. 2, and the registration, the holding of the iris data, and the authentication are performed by one device. You can do it.

  FIG. 5 is a diagram illustrating an appearance of a mobile phone with an authentication function as an example of the iris authentication device 15 in the present embodiment. The mobile phone with authentication function 20 shown in FIG. 5 is obtained by adding a camera 21 for capturing an iris image and a lighting 22 for capturing an iris to the mobile phone. In addition to the camera 21 and the lighting 22, a monitor 23, operation buttons 24, a speaker 25, a microphone 26, an antenna 27, and the like are provided. The illumination 22 is constituted by one or several near-infrared LEDs. A visible light cut filter is set in the camera 21 and receives only a near-infrared component. The monitor 23 displays an iris image being captured and an authentication result.

  In the present embodiment, the person to be authenticated performs iris authentication in a variety of environments, whether outdoors or indoors, or day or night, using a mobile phone with an authentication function as shown in FIG. And

  FIG. 6 is a flowchart showing the processing at the time of registration in the present embodiment. According to the flow of FIG. 6, processing at the time of registration in the personal authentication method according to the present embodiment will be described.

  First, the registrant uses the iris photographing device 16 to photograph a plurality of iris images having different pupil openings (SA1). In the iris photographing device 16, the photographing unit 32 can continuously photograph a plurality of iris images, and the light emission intensity of the illumination 31 can be controlled by the illumination control unit 33. Here, the illumination control unit 33 turns off the illumination 31 that was turned on at the start of imaging at a predetermined timing in a process in which the imaging unit 32 continuously captures a plurality of iris images of the registrant. .

  FIG. 7 is a graph showing changes in illumination control, pupil opening, and imaging timing when an iris image is captured. As shown in FIG. 7, in the present embodiment, it is assumed that the illumination 31 having a uniform light emission intensity is turned off at time t. Before time t, that is, when the illumination 31 is uniformly lit, the pupil opening is substantially constant. However, since the human pupil repeatedly contracts and expands slightly even under a constant brightness, the pupil opening slightly oscillates even in FIG. 7B.

  Then, at time t, when illumination 31 is turned off by the illumination control unit, the pupil enlarges until it adapts to the brightness, and after adaptation, repeats slight contraction and enlargement again. In this process, the photographing unit 32 continuously photographs a plurality of (here, 10) iris images at the photographing timing shown in FIG. 7B. The photographing timing is desirably several seconds to ten and several seconds from immediately before the time t when the light is turned off to when the pupil is completely dilated.

  As described above, by changing the illumination intensity of visible light, a plurality of iris images with various pupil openings can be captured. In addition, here, it is only necessary to turn off the light 31 which was turned on at the time of starting the photographing, so that complicated lighting control is not particularly required. Further, by changing the illuminance from bright to dark, the pain caused by the glare of the registrant can be reduced. Conversely, in the case of the control for changing the illumination from dark to bright, it becomes dazzling to the registrant, which is rather undesirable.

  That is, as an apparatus for photographing an iris image for personal authentication, illumination for performing visible light illumination around the face of the registered person, an illumination control unit for controlling the light emission intensity of the illumination, and an iris of the registered person A photographing unit for photographing an image, wherein the lighting control unit performs a predetermined timing of lighting that is turned on at the start of photographing in a process of photographing a plurality of iris images of the registered person continuously. It is preferable that the light is turned off at the same time.

  Accordingly, in the process of capturing the iris image, the illumination is merely changed from the on state to the off state, so that complicated illumination control is not required, and a plurality of iris images having different pupil opening degrees can be captured extremely easily. . Further, since the illuminance changes from light to dark, the registrant does not suffer excessive pain due to glare.

  Of course, instead of turning off the illumination, for example, the ambient brightness may be changed by capturing an iris image in a room having a window and closing the curtain at a certain time to block external light. Further, instead of simply turning on / off the lighting, the brightness may be changed in multiple steps.

  It is desirable that this iris photographing apparatus be installed in a room where external light does not enter. In such a place, when the illumination intensity is increased, an iris image in an extremely miotic state can be taken, and an iris image in an extremely mydriatic state can be taken after turning off the light because no external light enters. Further, an iris image between the miosis and the mydriasis can be captured in multiple stages.

Next, feature data is extracted from each of the plurality of iris images captured in step SA1 (SA2). In the iris registration device 14, the iris image acquisition unit 34 acquires an iris image photographed by the iris photographing device 16, and the feature extraction unit 35 extracts feature data from the acquired iris images. Here, it is assumed that the method described in Patent Document 2 (Japanese Patent Application Laid-Open No. 8-504979) is used. The outline of the iris authentication method of Patent Document 2 is as follows.
(1) The iris region is cut out by determining the outer edge of the iris (boundary between the iris and the sclera) and the outer edge of the pupil (the boundary between the pupil and the iris). (2) The cut out iris region is xy orthogonal. Convert from the coordinate system to the rθ polar coordinate system (3) Determine the analysis band (divide the radial direction into eight rings)
(4) Applying a multi-scale 2-d Gabor filter and binarizing the signal output from the Gabor filter as an iris code (5) Registered iris code registered in advance and iris code at the time of authentication (Exclusive OR) to calculate the Hamming distance between the two codes (6) If the Hamming distance is smaller than the threshold, accept as a person, otherwise reject as another person

  FIG. 8A is a diagram illustrating the positions of the outer edge of the iris and the outer edge of the pupil in FIG. 8A, and FIG. is there. At this point, the effects of the translation of the iris region are absorbed. FIG. 8C is a diagram expressing the iris region in the rθ polar coordinate system with the pupil center as the center (conversion of (2)). The actual pupil and iris edges are not exactly round. If both are intentionally approximated by a circle, the center of the pupil and the center of the iris are not concentric (eccentric). Difference in the degree of opening and the effect of scaling can be absorbed.

  FIG. 9 is a diagram illustrating the 8-ring analysis band determined in (3), and FIG. 10 is a diagram illustrating the iris code creation in (4). The luminance signal (a) after the analysis band in FIG. (B) shows a state in which (b) binarization is performed by applying a Gabor filter (c). Although the signal is actually a two-dimensional signal, it is shown here as one-dimensional for simplification of description. (A) is an angular direction luminance signal in one of the eight rings. Actually, a multi-scale Gabor filter is applied, and a single-scale Gabor filter has a real part and an imaginary part, but (b) and (c) show the results of applying a certain-scale Gabor filter real part. It is. The position of each bit in the iris code (c) after binarization based on the sign of the output of the Gabor filter can be associated with a certain position on the iris image.

  As described above, by performing the processes (1) to (6) on a plurality of iris images, the iris code which has absorbed the effects of translation, enlargement / reduction, difference in pupil opening, and pupil eccentricity can be obtained. Created as feature data.

Next, a pupil diameter / iris diameter ratio is calculated from each of the acquired iris images (SA3). This process is performed by the ratio calculator 36 of the iris registration device 14. Here, fitting is performed on the outer edge of the pupil with a circle having a radius r _{p and} the outer edge of the iris with a circle having a radius r _i , and the pupil diameter-iris diameter ratio R is calculated as
R = r _p / r _i
Is defined. The iris codes extracted from the plurality of iris images having different pupil opening degrees are respectively associated with the pupil diameter / iris diameter ratio R.

  In the present embodiment, the pupil diameter / iris diameter ratio is used as a “pupil opening degree index” for expressing the pupil opening degree. The reason is as follows. As a parameter expressing the degree of pupil opening, for example, “pupil diameter” itself can be used. However, in this case, it is necessary to satisfy the condition that the iris can always be photographed with a fixed size (the number of pixels). However, when the iris authentication function is added to a portable device such as a mobile phone or a PDA, the photographing person performs the photographing while holding the portable device equipped with the camera with his / her hand. There is no guarantee that the distance is constant, and it is likely to fluctuate. For this reason, since the size itself of the iris region changes on the iris image, the pupil opening cannot be accurately represented only by the pupil diameter. On the other hand, when the pupil diameter / iris diameter ratio is used as the pupil opening index as in the present embodiment, even if the size of the iris region is not constant on the image, the pupil opening is accurately calculated. Can be expressed.

  Further, as another pupil opening degree index, ambient brightness at the time of imaging may be used. As the surrounding brightness, for example, illuminance (unit: lx) can be used. Since the human pupil contracts or expands according to the surrounding brightness, the approximate pupil opening can be expressed by the illuminance value. It is desirable that the sensor position for measuring the illuminance is near the eye position of the subject, but if it is difficult to install the sensor, another position such as a built-in device for photographing may be used.

  Next, feature data to be finally stored is selected (SA4). In the present embodiment, as shown in FIG. 7, ten iris images are shot for a certain registrant. All of the iris codes obtained for each of the ten iris images may be registered in the iris database 12, but here, in order to minimize the amount of data stored in the iris database 12, a minimum necessary iris code is set. Shall be selected.

  The selection of the iris code is specifically performed by the following processing. First, the iris codes are rearranged so that the pupil diameter-iris diameter ratio R is in ascending order (or descending order), and the iris codes are numbered. Next, a Hamming distance between different iris codes is calculated as a comparison between the feature data. The calculation method here is the same as the calculation of the Hamming distance in the processing at the time of authentication described later.

In the present embodiment, since there are 10 iris codes, 45 hamming distances (= ₁₀ C ₂ ) are calculated. As a result, a Hamming distance matrix as shown in FIG. 11 is obtained. Strictly speaking, the iris pattern does not contract uniformly, so that the closer the iris code to the value R of the pupil diameter / iris diameter ratio, the smaller the hamming distance, and the farther the iris code, the greater the hamming distance.

  Then, for each iris code, it is determined whether or not the Hamming distance between two adjacent iris codes is equal to or greater than a predetermined threshold. If both of the hamming distances are less than the threshold, the iris code can be omitted. Is determined. The predetermined threshold is, for example, 0.20.

  Specifically, focusing on the iris code 1 in FIG. 11, the Hamming distance with the iris code 0 is 0.15, and the Hamming distance with the iris code 2 is 0.16. That is, since both the Hamming distances are less than the threshold value, that is, less than 0.20, it is determined that the iris code 1 can be omitted. In this determination, even if the iris code at the time of authentication is close to the iris code 1, since the authentication is successful by comparing with the iris code 0 or the iris code 2, the iris code 1 may be omitted. , Based on the idea.

  Then, in FIG. 11, the rows and columns of the iris code 1 are deleted, and the same operation is repeated. Focusing on the iris code 2, since the Hamming distance from the iris code 0 is 0.20, which is not less than the threshold value, it is determined that the iris code 2 cannot be omitted. Next, when focusing on the iris code 3, the hamming distance with the iris code 2 is 0.14, and the hamming distance with the iris code 4 is 0.13. Is determined to be omissible. By repeating such processing, six iris codes 0, 2, 4, 6, 8, and 9 are finally selected.

  As described above, the iris code to be finally registered is determined with reference to the Hamming distance between the iris codes, whereby the data amount of the iris database 12 can be reduced.

  As a method for selecting feature data to be stored in the iris database, various methods can be considered in addition to using the Hamming distance of the iris code. For example, the pupil diameter / iris diameter ratio may be used. For example, the pupil diameter / iris diameter ratio may be in the range of 0.1 to 0.3, 0.3 to 0.5, and 0.5 to 0.7. One of them may be selected. In addition, whether or not the degree of spread of the eyes is sufficiently large, whether or not the image is blurred due to blinking or movement of the subject, whether or not the focus value is correct, and the like can be used as selection criteria.

  The selection of the iris code can also be performed by the following method. That is, when it is considered that any one of the plurality of extracted iris codes is actually registered, a graph as shown in FIG. 12 can be created from the Hamming distance between the temporary registration code and each iris code. In FIG. 12, the vertical axis is the Hamming distance, the horizontal axis is the pupil diameter / iris diameter ratio, and each point corresponds to each iris code other than the temporary registration code.

  Then, from the graph of FIG. 12, it is possible to obtain a range A1 of the pupil diameter / iris diameter ratio that satisfies an acceptable FRR (rejection rate, for example, 1%). The same processing is performed for other iris codes to determine the range of the pupil diameter / iris diameter ratio that satisfies the allowable FRR. Then, by selecting the iris code so that the entire range of the pupil diameter / iris diameter ratio can be covered, the necessary minimum registered iris code can be determined.

  However, a necessary condition for performing this method is that the number of iris data obtained at the time of registration is sufficiently large. In addition, at the time of authentication, it is conceivable that the individual may be rejected due to other factors (eg, camera defocus, concealment by eyelids / eyelashes, etc.). It is desirable that they overlap little by little.

  As a result of experiments by the inventors of the present application, for one registrant, if at least three pieces of feature data obtained from the iris images in the miotic state, the normal state, and the mydriatic state are registered, at least Registrants have proven to be well authenticated. Here, the miotic state is when the pupil diameter / iris diameter ratio is 0.1 to 0.3, and the normal state is when the pupil diameter / iris diameter is 0.3 to 0.5 and the mydriatic state. Means that the pupil diameter / iris diameter ratio is 0.5 to 0.7.

  Finally, the iris code as the feature data selected in step SA4 is stored in the iris database 12 together with the pupil diameter / iris diameter ratio as the pupil opening degree index (SA5). Here, the feature transmitting unit 37 of the iris registration device 14 transmits the iris code and the pupil diameter / iris diameter ratio to the iris database 12 via the network 13.

  FIG. 13 is a diagram schematically showing an example of the data content of the iris database 12, in which a total of N iris codes are registered for a plurality of (M) registrants. As shown in FIG. 13, a pupil diameter / iris diameter ratio is attached to each iris code as additional information.

  FIG. 14 is a flowchart showing a process at the time of authentication in the present embodiment. The processing at the time of authentication in the iris authentication method according to the present embodiment will be described with reference to the flow of FIG.

  First, the person to be authenticated captures an iris image by the iris image acquisition unit 41 of the iris authentication device 15 (SB1). Here, when a portable device such as the mobile phone 20 with an authentication function shown in FIG. 5 is used as the iris authentication device 15, authentication can be performed regardless of place and time. In the daytime outdoors, an iris image in a miotic state is taken, and at night or in a dark place, an iris image in a mydriatic state is taken.

  Next, the feature extraction unit 42 extracts an iris code as feature data from the captured iris image (SB2), and the ratio calculation unit 43 calculates the pupil diameter / iris diameter ratio from the captured iris image as a pupil opening index. Calculate (SB3). The processing in steps SB2 and SB3 is the same as that in steps SA2 and SA3 at the time of registration, and a description thereof will not be repeated.

  Next, the feature data acquisition unit 44 selects an iris code to be collated from the iris database 12 by comparing the pupil diameter / iris diameter ratio (SB4). Here, it is assumed that the iris code registered in addition to the pupil diameter / iris diameter ratio that is close to the pupil diameter / iris diameter ratio at the time of authentication is registered.

  For example, as shown in FIG. 1, when the pupil diameter / iris diameter ratio R in the iris image at the time of authentication is 0.41, an iris code whose ratio R is close to 0.41 is acquired from the iris database 12 as a collation target. I do. FIG. 1 is a diagram for explaining one-to-N authentication without declaring its own ID. In the case of one-to-N authentication, as shown in FIG. 1, for each registrant registered in the iris database 12, one iris code having a ratio R closest to 0.41 is acquired. On the other hand, in the case of one-to-one authentication in which the user ID is declared, one iris code whose ratio R is closest to 0.41 is acquired for the registrant corresponding to the declared ID.

  Then, the authentication unit 45 performs authentication by comparing the iris code extracted by the feature extraction unit 42 with the iris code acquired by the feature data acquisition unit 44 (SB5). Here, similarly to the method of Patent Document 2, the comparison is performed while performing the relative shift of the iris code in order to compensate for the influence of the inclination of the face of the target person and the rotation of the eyeball. The relative shift at the time of the code comparison is performed by repeating the comparison process while shifting the iris codes in the angular direction with respect to each other as if the iris codes shown in FIG. Be executed. Then, the hamming distance at the time of the shift when the hamming distance becomes the minimum is output as the final hamming distance. Authentication is performed based on whether or not this final Hamming distance is equal to or greater than a threshold. If the final Hamming distance is equal to or greater than the threshold, the authenticated person is determined to be another person, while if less than the threshold, the authenticated person is determined to be the person. .

  Here, an example in which authentication is performed by the iris authentication device 15 has been described, but instead, the iris authentication device, which is a terminal owned by the user, performs processes (SB1 to SB3) up to feature extraction and ratio calculation. The extracted feature data and the value of the pupil diameter / iris diameter ratio may be transmitted to the iris authentication server 11, and the iris authentication server 11 may perform feature data acquisition and authentication (SB4, SB5).

  As described above, according to the present embodiment, at the time of registration, a plurality of iris images having different pupil opening degrees are obtained for the registrant, a plurality of iris codes are extracted from the plurality of iris images, and the pupil diameter / iris is extracted. The iris code is registered in association with the diameter ratio. At the time of authentication, the iris code to be collated is selected by comparing the pupil diameter / iris diameter ratio from the plurality of registered iris codes. That is, since the iris code extracted from the iris image captured at the time of authentication and the iris image having the same degree of pupil opening is selected as a collation target, the pupil at the time of authentication is in the mydriatic state even if it is in the miotic state. Even in this case, it is possible to execute personal authentication with a reduced rejection rate.

  Also, since the iris codes to be compared are narrowed down using the pupil diameter / iris diameter ratio, a plurality of iris codes having different pupil diameter / iris diameter ratios are simply registered, and the iris code having the smallest hamming distance at the time of authentication is registered. The collation time can be greatly reduced as compared with the case where is selected.

(Modification of First Embodiment)
In the first embodiment described above, at the time of registration, a plurality of iris images having different pupil opening degrees are acquired for a registered person, and a plurality of feature data respectively obtained from the plurality of iris images are registered in the iris database. It was taken. Here, a modified example thereof will be described.

  FIG. 22 is a flowchart showing the processing at the time of registration in this modification. According to the flow of FIG. 22, processing at the time of registration in the personal authentication method according to the present modification will be described.

  First, the registrant captures an iris image (SA1 '). The difference from the first embodiment is that it is not always necessary to photograph a plurality of images, and it is not necessary to use a device capable of controlling the intensity of visible light illumination such as the iris photographing device 16. Here, it is assumed that registration is performed using a portable device as shown in FIG. 5, for example.

  Next, an iris code as feature data is extracted from the iris image captured in step SA1 '(SA2'). The pupil diameter / iris diameter ratio is calculated from the acquired iris image (SA3 '). The method of extracting the iris code from the iris image and the method of calculating the pupil diameter / iris diameter ratio from the iris image are the same as those in the first embodiment, and thus description thereof will be omitted.

Finally, the iris code as the feature data is stored in the iris database 12 together with the pupil diameter / iris diameter ratio as the pupil opening index (SA4 ').
At this time, the iris registration device 14 or the iris authentication server 11 may check whether the iris code has already been registered for the same registrant. For example, if the iris code has already been registered, the pupil diameter / iris diameter ratio between the registered iris code and the new iris code is compared with each other. Do not save the iris code. This is because, even if there are a plurality of registered iris codes having similar pupil diameter / iris diameter ratios, they do not contribute much to the improvement of the authentication performance, but merely increase the capacity of the iris database. The determination as to whether or not the values of the pupil diameter / iris diameter ratio are close may be made using a predetermined threshold.

  The registrant may perform the registration processing of steps SA1 'to SA4' continuously or a plurality of times at intervals. In that case, by performing registration processing in an environment with different ambient brightness, iris images with different pupil opening degrees can be acquired, so that a plurality of iris codes having different values of the pupil diameter / iris diameter ratio can be registered. .

  On the other hand, the processing at the time of authentication according to the present modification is the same as that of the first embodiment, and the description will be made according to the flow of FIG.

  The processing in steps SB1 to SB3 is the same as in the first embodiment, and a description thereof will not be repeated.

  Next, the feature data acquisition unit 44 specifies the iris code to be collated from the iris database 12 by comparing the pupil diameter / iris diameter ratio (SB4). Here, it is assumed that the iris code registered in addition to the pupil diameter / iris diameter ratio that is close to the pupil diameter / iris diameter ratio at the time of authentication is registered.

  For example, as shown in FIG. 23, when the pupil diameter / iris diameter ratio R in the iris image at the time of authentication is 0.25, an iris code having a ratio R close to 0.25 is obtained from the iris database 12 as a collation target. I do. FIG. 23 shows one-to-N authentication without declaring its own ID. In the case of one-to-N authentication, as shown in FIG. 23, for each of the registrants registered in the iris database 12, one iris code whose ratio R is closest to 0.25 is acquired. On the other hand, in the case of one-to-one authentication in which the user ID is declared, one iris code whose ratio R is closest to 0.25 is acquired for the registrant corresponding to the declared ID.

  Then, the authentication unit 45 performs authentication by comparing the iris code extracted by the feature extraction unit 42 with the iris code acquired by the feature data acquisition unit 44 (SB5). The processing in step SB5 is the same as in the first embodiment, and a description thereof will not be repeated.

In the present modification, by assigning the pupil diameter / iris diameter ratio to the iris code to be registered at the time of registration, it is possible to specify the iris code having the closest pupil diameter / iris diameter ratio as the matching target at the time of authentication. Therefore, authentication can be performed such that the personal rejection rate becomes smaller. However, in the present modified example, unlike the first embodiment, at the time of registration, a plurality of iris images having different pupil openings are not photographed by changing the brightness. For this reason, as shown in FIG. 23, the pupil diameter-iris diameter ratio of the registered iris codes may not have close intervals. For this reason, there is a possibility that the false rejection rate will be higher than in the first embodiment.
In the case of one-to-one authentication, an authentication process as shown in FIG. 24 can be performed so as not to increase the personal rejection rate. The processing in steps SB1 to SB3 is the same as in the first embodiment, and a description thereof will not be repeated.

  Next, the feature data acquisition unit 44 searches the iris database 12 for an iris code to be matched by comparing the pupil diameter / iris diameter ratio (SB8). Here, it is assumed that the iris code registered in addition to the pupil diameter / iris diameter ratio which is close to the pupil diameter / iris diameter ratio at the time of authentication by a predetermined degree is searched.

  For example, as shown in FIG. 25, when the pupil diameter / iris diameter ratio R in the iris image at the time of authentication is 0.25, the ratio R in the iris database 12 is within a predetermined range (here, for example, An iris code within ± 0.15) is searched for a collation target. FIG. 25 shows one-to-one authentication for declaring one's own ID. If the declared ID is i, the iris code of the registrant i is searched for the iris code whose ratio R is within the range of 0.25 ± 0.15 and which is closest to 0.25.

  It is checked whether or not the corresponding iris code exists (SB9). If the iris code exists (Yes), authentication is performed using the iris code as a collation target (SB10). The processing in step SB10 is the same as step SB5 in the first embodiment, and a description thereof will not be repeated. On the other hand, when the corresponding iris code does not exist (No), the authentication process is stopped (SB11). In the example of FIG. 25, since there is no corresponding iris code, the process proceeds to step SB11.

  Here, in step SB11, not only the authentication processing is simply stopped, but also advice to the person to be authenticated for reducing the false rejection rate can be given. That is, based on the pupil opening index at the time of authentication and the pupil opening index related to the registered feature data, a preferable iris image photographing condition is estimated, and the iris image is re-acquired under the estimated photographing condition. Prompt the certifier. The reacquisition of the iris image here may be for re-authentication or registration of new feature data. This reduces the false rejection rate and improves user convenience. The presentation of the advice may be performed, for example, via the monitor 23 of the mobile phone 20 in FIG.

  For example, in the case of FIG. 25, for the registrant i, the minimum value of the pupil diameter / iris diameter ratio of the registered iris code is 0.43, and the pupil diameter / iris diameter ratio at the time of authentication is 0.25. In the authentication of (1), the user is prompted to take a picture in a darker place so that the pupil diameter / iris diameter ratio becomes larger than 0.25. Alternatively, for future authentication, registration in a bright place may be prompted so that an iris code having a pupil diameter-iris diameter ratio smaller than 0.43 can be registered.

(Second embodiment)
In the second embodiment of the present invention, at the time of registration, a relational expression between the feature data and the pupil opening index is obtained, and parameters for expressing the relational expression are registered. Then, at the time of authentication, the extracted pupil opening degree index is substituted into a relational expression represented by the registered parameters to obtain feature data to be collated.

  The personal authentication method according to the present embodiment is realized, for example, in the authentication system of FIG. FIG. 15 is a block diagram illustrating the configuration of the iris registration device 14A according to the present embodiment, and the same reference numerals as in FIG. 3 denote the same components as in FIG. The relational expression calculation unit 38 calculates a relational expression between the characteristic data and the pupil diameter / iris diameter ratio as a pupil opening degree index, and obtains a parameter representing this relational expression. The feature transmitting unit 37 transmits the parameters obtained by the relational expression calculating unit 38 to the iris database 12.

  FIG. 16 is a block diagram showing the configuration of the iris authentication device 15A according to the present embodiment, and the same reference numerals as in FIG. 4 denote the same components as in FIG. The characteristic data calculation unit 46 acquires characteristic data to be collated by substituting the pupil diameter / iris diameter ratio at the time of authentication into the relational expression represented by the parameters registered in the iris database 12.

  FIG. 17 is a flowchart showing the processing at the time of registration in the present embodiment. The processing at the time of registration in the personal authentication method according to the present embodiment will be described with reference to the flow of FIG.

  The processing of steps SA1 to SA3 is the same as in the first embodiment. That is, a plurality of iris images having different pupil openings are photographed by the iris photographing device 16 for the registered person (SA1), and the feature extraction unit 35 of the iris registration device 14A extracts iris as feature data from the plurality of iris images. The codes are extracted (SA2), and the pupil diameter / iris diameter ratio is calculated from the plurality of iris images by the ratio calculation unit 36 of the iris registration device 14A (SA3). However, in step SA2, for each iris image, the signal after applying the Gabor filter to the image represented by the polar coordinates as shown in FIG. 10B is stored together with the iris code.

  Next, the relational expression calculation unit 38 of the iris registration device 14A calculates a relational expression between the pupil diameter / iris diameter ratio and the characteristic data for each dimension (SA6). FIG. 18 is a flowchart showing details of the processing in step SA6.

First, the Hamming distance between the i-th code and the j-th code (j ≠ i) is calculated, based on the i-th code, for the plurality (K) of iris codes created in step SA2 ( SA61). The method of calculating the Hamming distance between codes is the same as the method described in the first embodiment. At this time, the shift amount (rotation angle θ _j ) when the hamming distance is calculated is stored together with the hamming distance (SA62).

Then, each of the output signals of the Gabor filter which have been stored at step SA2, is rotated by the rotation angle θ _j (SA63). Finally, the value at the same position in each output signal is observed, and the relationship between the value of the pupil diameter / iris diameter ratio and the output signal value is fitted by a polynomial (D-order) (SA64).

  FIG. 19 is a diagram for conceptually explaining the processing of step SA6. In the figure, (a) is a plurality of iris images having different pupil openings taken in step SA1, and the iris region is simply represented by a double circle of the outer edge of the iris and the outer edge of the pupil. Also, the value of the pupil diameter / iris diameter ratio R calculated in step SA3 is shown. FIG. 19 (b) shows the output of the Gabor filter stored in step SA2 shifted by the shift amount corresponding to the Hamming distance calculated in step SA61, and FIG. 19 (c) shows the Gabor filter in FIG. 19 (b). It is a figure which shows the result of having plotted the value in the coordinate (i, j) among the filter outputs on the graph which sets a horizontal axis to a pupil diameter-iris diameter ratio, and fitting it by D following formula.

Although FIG. 19C shows the Gabor filter output of the real part (or the imaginary part) at a certain coordinate, this processing is actually performed for all the coordinates for which the characteristic data is obtained, for the real part and the imaginary part. The analysis is performed for each of the analysis frequencies of the Gabor filter. For example, when a Gabor filter having S points in the radial direction, T points in the angular direction, and F types of frequencies is applied to the polar coordinate image, Z such D-order expressions are created.
Z = S × T × F × 2
Multiplying by 2 corresponds to the real part and the imaginary part.

If the number of dimensions D is set to be small, the data amount of the iris database 12 is reduced, but the fitting accuracy is reduced. Conversely, if the setting is large, the storage capacity of the iris database 12 increases, but the accuracy of fitting improves. In the present embodiment, “3” is used as the number of dimensions D. When D = 3, the relational expression is
Y = aX ³ + bX ² + cX + d
It becomes. Here, X is a pupil diameter / iris diameter ratio, Y is a Gabor filter output at a certain position on polar coordinates, and a, b, c, and d are coefficients.

  The reason for choosing D = 3 is as follows. When iris authentication is used indoors, the change in iris pattern can be approximated as a rubber sheet that expands and contracts uniformly. The fact that it can be approximated by a rubber sheet means that the sign of the Gabor filter output does not change even if the pupil diameter / iris diameter ratio changes. If the sign does not change, the bits of the iris code do not change. That is, in a range close to the pupil diameter-iris diameter ratio of the registered iris code, Y can be represented by a relatively low-order expression of X (D = 1 or 2). On the other hand, when the pupil diameter / iris diameter ratio of the registered iris code deviates from the pupil diameter / iris diameter ratio, the bits of the iris code are inverted, and the sign of the output of the Gabor filter changes. In order to express such a relationship, a higher-order equation is required. However, as the dimension increases, the storage capacity for storing the coefficients increases. Therefore, in the present embodiment, D = 3 is selected. Of course, the value of D may be set to be larger than 3 in a case where accuracy is prioritized even when the storage capacity increases.

  The coefficients of the relational expression thus calculated are stored as parameters in the iris database 12 (SA7). FIG. 20 shows an example of the data content of the iris database 12, in which coefficients a, b, c, and d of the relational expression are stored. The first suffix of each coefficient corresponds to the number of the feature data, and the second suffix corresponds to the number of dimensions.

  Here, a method has been described in which the relational expression between the pupil diameter / iris diameter ratio and the Gabor filter output is expressed by polynomial fitting, but in addition to this, it is expressed by an orthogonal basis such as cosine transform. Is also good.

Here, a storage capacity for storing feature data will be described. In the method of Patent Document 2, feature data is represented by 2048-dimensional binary information, and a capacity of 2048 bits (= 256 bytes) is required for each piece of feature data. On the other hand, in the case of the present embodiment, when each dimension is represented by a cubic expression and each coefficient is represented by a single precision floating point (float; 4 bytes), the capacity per feature data is:
2048 × 4 × 4 = 32768 (bytes)
Thus, a capacity 128 times larger than that of Patent Document 2 is required.

  When it is desired to reduce the capacity of the iris database, each parameter may be thinned out and registered. When the 2-D Gabor filter is applied in the angular direction and the radial direction in the two-dimensional polar coordinates, the output values after the convolution have a correlation at spatially close positions in the angular direction and the radial direction. Also in FIG. 10B, since the values are continuous when viewed locally, it can be seen that there is a correlation at a position close to the angular direction. Therefore, it is assumed that each coefficient value of the following equation obtained from the Gabor filter output value has a correlation, and the coefficient is registered at every several points in both the angular direction and the radial direction. If coefficients are registered at every other point in both the angular direction and the radial direction, the capacity increase of 128 times in the above example can be suppressed to 128/2/2 = 32 times. At the time of authentication, a coefficient value can be calculated by interpolating from surrounding coefficient values.

  FIG. 21 is a flowchart showing a process at the time of authentication in the present embodiment. The processing at the time of authentication in the personal authentication method according to the present embodiment will be described with reference to the flow of FIG.

  The processing of steps SB1 to SB3 is the same as in the first embodiment. That is, first, the person to be authenticated captures an iris image by the iris image acquisition unit 41 of the iris authentication device 15A (SB1). Next, the feature extraction unit 42 extracts an iris code as feature data from the captured iris image (SB2), and the ratio calculation unit 43 calculates the pupil diameter / iris diameter ratio from the captured iris image as a pupil opening index. Calculate (SB3).

  Next, the feature data calculation unit 46 obtains a relational expression from the parameters registered in the iris database 12, substitutes the pupil diameter / iris diameter ratio calculated by the ratio calculation unit 43 into this relational expression, Is calculated (SB6). For example, when the pupil diameter / iris diameter ratio extracted from the iris image at the time of authentication is 0.41, the pupil diameter / iris diameter ratio is expressed by the following equation created from the parameters stored in the iris database as shown in FIG. And substitute 0.41 and calculate the output value. Then, the output value is binarized. This process is performed for all dimensions to generate a binary iris code.

  At this time, in the case of one-to-N authentication, an iris code is acquired for each of the registrants registered in the iris database 12. On the other hand, in the case of one-to-one authentication in which the user's ID is declared, the pupil diameter / iris diameter ratio is substituted for the registrant corresponding to the declared ID into a D-order equation created from registered parameters. Acquire one iris code.

  The authentication unit 45 performs authentication by comparing the iris code extracted by the feature extraction unit 42 with the iris code obtained by the feature data calculation unit 46 (SB7). The processing here is the same as step SB5 in the first embodiment, and a description thereof will not be repeated.

  As described above, according to the present embodiment, at the time of registration, a plurality of iris images having different pupil opening degrees are obtained for the registrant, and parameters of the following equation for generating feature data from the plurality of iris images are calculated. At the time of authentication, the ratio of the pupil diameter to the iris diameter at the time of authentication is substituted into the following equation D calculated from the registered parameters to determine the iris code to be compared. That is, since the iris code corresponding to the pupil opening of the iris image captured at the time of authentication is generated as a target to be compared, the pupil at the time of authentication is rejected even if the pupil is in the miotic state or the mydriatic state. Personal authentication with a reduced rate can be executed.

  In Patent Document 3, instead of absorbing the difference between the pupil diameters at the time of registration and at the time of authentication by using a polar coordinate expression as in Patent Document 2, a plurality of registered images having different pupil diameters are prepared. The difference between the pupil diameter at the time of authentication and at the time of authentication is absorbed. Therefore, in order to increase the authentication accuracy, a large number of iris images or feature data having slightly different pupil diameters are required, which leads to an increase in the capacity of the database. This is because if the pupil diameter is slightly different, different feature data will be obtained. In addition, there is a problem that the authentication time is significantly increased because it is necessary to collate with many iris images at the time of authentication. In the case of one-to-N authentication, a large number of feature data must be registered for a large number of registrants, and all of them must be collated at the time of authentication, so this problem becomes more serious.

  On the other hand, in each of the above-described embodiments, a plurality of iris codes are registered in association with the pupil diameter / iris diameter ratio, and the iris code to be collated is compared by comparing the pupil diameter / iris diameter ratio at the time of authentication. Or register the parameters of the following equation to generate the feature data, and substitute the pupil diameter / iris diameter ratio at the time of authentication into the following equation calculated from the registered parameters for verification. Since the target iris code is determined, the capacity of the iris database and the authentication time do not increase, and therefore, it can be said that the practicality is significantly improved as compared with Patent Document 3.

  In addition, at the time of authentication outdoors, not only a change in the iris pattern but also reflection of external light can be a factor inhibiting authentication. This reflection is dealt with, for example, by Japanese Patent Application No. 2002-28446 by the present inventors. It is possible.

(Third embodiment)
In the third embodiment of the present invention, at the time of registration, together with an iris code having a pupil diameter / iris diameter ratio, a conversion rule from the registered iris code to another iris code having a different pupil diameter / iris diameter ratio is set. , Register with the iris database. Then, at the time of authentication, an iris code to be collated is generated based on the iris code registered in the iris database and the conversion rule using the determined pupil diameter / iris diameter ratio.
FIG. 26 is a flowchart showing the processing at the time of registration in the present embodiment. The processing at the time of registration in the personal authentication method according to the present embodiment will be described with reference to the flow of FIG.

  First, the registrant captures a plurality of iris images having different pupil openings (SA1). Then, feature data is respectively extracted from the plurality of iris images captured in step SA1 (SA2), and the pupil diameter-iris diameter ratio is calculated (SA3). Since these processes are the same as those in the first embodiment, the description is omitted.

  Next, at least one feature data to be registered in the iris database is specified (SA8), and the specified feature data is stored in the iris database (SA9). Here, it is assumed that feature data having the most average pupil diameter / iris diameter ratio is selected as registered feature data. The reason is that it is expected that the probability of inputting an iris image having an average pupil diameter / iris diameter ratio during authentication will be the highest. For example, the maximum value and the minimum value are obtained from the pupil diameter / iris diameter ratio of the plurality of acquired iris images, and the pupil diameter / iris diameter ratio closest to the average value ((maximum value + minimum value) / 2) is obtained. Select feature data. When the pupil diameter / iris diameter ratio of the registered iris data is close to the pupil diameter / iris diameter ratio at the time of authentication, authentication can be performed with a reduced rejection rate without converting the registered iris data.

  Further, feature data having a pupil diameter / iris diameter ratio of an iris image captured in a situation where the registrant performs the authentication operation most frequently may be selected as the registered feature data. For example, for a person who frequently authenticates indoors, feature data having a pupil diameter / iris diameter ratio at indoor brightness is selected, and for a person who frequently authenticates outdoors, pupil diameter at outdoor brightness. -Select feature data having an iris diameter ratio.

  Next, a conversion rule from the registered feature data stored in step SA9 to feature data having another pupil diameter / iris diameter ratio is calculated (SA10). Strictly speaking, when the pupil diameter changes, the iris pattern contracts unevenly in the circumferential direction. On the other hand, as can be seen from FIG. 23, even if the pupil diameter fluctuates somewhat, the change in the hamming distance is small, so that if the difference between the pupil diameter and the iris diameter ratio between registration and authentication is small, authentication is possible. Here, the conversion rule of the feature data is calculated when the difference between the pupil diameter and the iris diameter ratio is large to some extent.

  Here, the pupil diameter / iris diameter ratio of the registered feature data is R0, and the pupil diameter / iris diameter ratio of the feature data for which the conversion rule is calculated is R1. As shown in FIG. 27, a part of the iris code having the pupil diameter / iris diameter ratio R0 is found to correspond to a part of the iris code having the pupil diameter / iris diameter ratio R1. The conversion rule is expressed by the amount of movement (bit shift amount) between them. This bit shift amount basically increases as the difference between the pupil diameter and the iris diameter ratio of the two codes increases. In the example of FIG. 27, the conversion rule is expressed as “shift the 6th to 10th bits by 3 bits” and “shift the 19th to 25th bits by −3 bits”.

  Then, the calculated conversion rule is stored in the iris database (SA11). The conversion rules are stored, for example, in a table format as shown in FIG. Here, the conversion rule is stored in the iris database. However, if the user always uses the same terminal such as the portable authentication device in FIG. 5, the conversion rule may be stored in the authentication device. I do not care. In that case, only the table corresponding to the eye of the person to be authenticated is stored in the authentication device.

  FIG. 29 is a flowchart showing a process at the time of authentication in the present embodiment. The processing at the time of authentication in the personal authentication method according to the present embodiment will be described with reference to the flow of FIG.

  The processing in steps SB1 to SB3 is the same as in the first embodiment, and a description thereof will be omitted.

  Next, based on the conversion rule, the registered feature data is converted using the pupil diameter / iris diameter ratio at the time of authentication (SB12). When feature data of a plurality of persons (eyes) is registered in the iris database, conversion rules are also stored for each person (eye). In the case of one-to-N authentication, the registered iris code is converted using the pupil diameter / iris diameter ratio of the iris image acquired at the time of authentication based on the conversion rule corresponding to the registrant of interest. In the case of one-to-one authentication, the registered iris code is converted using the pupil diameter-iris diameter ratio of the iris image acquired at the time of authentication, based on the conversion rule corresponding to the person to be authenticated. When the conversion rule is stored in the authentication device, all the registered iris codes may be converted using the conversion rule of the person to be authenticated stored in the authentication device at the time of one-to-N authentication.

  Then, authentication is performed by comparing the matching target feature data obtained by converting the registered feature data with the feature data at the time of authentication (SB13).

  As described above, according to the present embodiment, at the time of registration, along with the iris code, a conversion rule to an iris code having a different pupil diameter / iris diameter ratio is registered, and at the time of authentication, it is acquired at the time of authentication based on the conversion rule. The registered iris code is converted using the pupil diameter / iris diameter ratio of the obtained iris image to generate a collation target. This makes it possible to perform authentication while suppressing an increase in the personal rejection rate.

  In the above embodiments, the iris code is used as the feature data of the iris image. However, other feature data may be used. Although the polar coordinate system expression is used for generating the feature data, this need not be used.

  Note that the processing described in each of the above embodiments can be implemented in whole or in part by causing a computer to execute a program. For example, a computer and a memory as a recording medium for storing an iris authentication program are mounted on an iris authentication device. The iris authentication program may be any program that causes a computer to execute the processing at the time of authentication described in each of the above embodiments.

  INDUSTRIAL APPLICABILITY The present invention enables sufficiently accurate personal authentication with a reduced rejection rate in various situations. For example, mobile phones and PDAs are equipped with an iris authentication function and can be used for mobile applications for e-commerce. It is effective when performing authentication.

FIG. 3 is a diagram conceptually showing a personal authentication method according to the first embodiment of the present invention. It is a figure showing the whole iris authentication system composition in each embodiment of the present invention. It is a block diagram showing composition of an iris registration device and an iris photography device in a 1st embodiment of the present invention. FIG. 1 is a diagram illustrating a configuration of an iris authentication device according to a first embodiment of the present invention. FIG. 1 is a diagram illustrating an appearance of a mobile phone with an authentication function as an example of an iris authentication device. 5 is a flowchart illustrating a registration process according to the first embodiment of the present invention. 9 is a graph showing changes in illumination control and pupil opening during imaging of an iris image, and imaging timing. (A) is a diagram showing an iris region, (b) is a diagram showing the iris region in an xy coordinate system, and (c) is a diagram showing the iris region in a polar coordinate system. It is a figure showing an analysis band. It is a figure showing creation of an iris code. It is a matrix of the Hamming distance between iris codes. It is the graph which plotted the Hamming distance between iris codes. 5 is an example of data contents of an iris database according to the first embodiment of the present invention. 5 is a flowchart illustrating a process at the time of authentication according to the first embodiment of the present invention. It is a block diagram showing the composition of the iris registration device in a 2nd embodiment of the present invention. It is a block diagram showing the composition of the iris authentication device in a 2nd embodiment of the present invention. It is a flow chart which shows processing at the time of registration in a 2nd embodiment of the present invention. It is a flowchart which shows the detail of the process of step SA6 in the flow of FIG. FIG. 18 is a diagram conceptually illustrating the process of step SA6 in the flow of FIG. 17. It is an example of the data content of the iris database 12 in 2nd Embodiment of this invention. It is a flow chart which shows processing at the time of attestation in a 2nd embodiment of the present invention. It is a flowchart which shows the process at the time of registration in the modification of 1st Embodiment of this invention. It is a figure which shows notionally the process at the time of the authentication in the modification of 1st Embodiment of this invention. It is a flow chart which shows processing at the time of attestation in a modification of a 1st embodiment of the present invention. It is a figure which shows notionally the process at the time of the authentication in the modification of 1st Embodiment of this invention. It is a flow chart which shows processing at the time of registration in a 3rd embodiment of the present invention. It is a figure which shows notionally an example of the calculation method of the conversion rule in 3rd Embodiment of this invention. It is an example of the conversion rule stored in the iris database. It is a flow chart which shows processing at the time of attestation in a 3rd embodiment of the present invention.

Explanation of reference numerals

Reference Signs List 11 iris authentication server 12 iris database 13 network 14, 14A iris registration device 15, 15A iris authentication device 16 iris imaging device 31 illumination 32 imaging unit 33 illumination control unit 34 iris image acquisition unit 35 feature extraction unit 36 ratio calculation unit 37 Transmission unit 38 Relational expression calculation unit 41 Iris image acquisition unit 42 Feature acquisition unit 43 Ratio calculation unit 44 Feature data acquisition unit 45 Authentication unit 46 Feature data calculation unit

Claims (17)

A personal authentication method using an iris image,
At the time of registration,
Obtain an iris image for the registrant,
From the acquired iris image, feature data and a pupil opening index are obtained,
Using the obtained feature data and the pupil opening index, register data on the registrant in the iris database,
At the time of authentication,
Obtain an iris image for the subject,
From the acquired iris image, feature data and a pupil opening index are obtained,
Referring to the data registered for the registrant in the iris database, using the pupil opening index obtained at the time of authentication, to obtain feature data to be collated,
A personal authentication method characterized by determining whether or not the person to be authenticated is the same as the registrant by comparing the characteristic data to be collated with the characteristic data obtained at the time of authentication. In claim 1,
At the time of registration,
The feature data is registered in the iris database in association with the registrant, together with the pupil opening degree index,
At the time of authentication,
From the feature data registered in association with the registrant in the iris database, by comparing the pupil opening index registered together with the feature data and the pupil opening index obtained at the time of authentication, A personal authentication method characterized by specifying the characteristic data to be collated. In claim 2,
At the time of registration, a personal authentication method characterized by registering at least three pieces of feature data obtained from an iris image in a miotic state, a normal state, and a mydriatic state for the registered person. In claim 2,
At the time of registration,
For the registrant, obtain a plurality of iris images having different pupil openings,
From each of the plurality of acquired iris images, feature data is obtained,
A personal authentication method characterized by selecting feature data to be registered in the iris database from the plurality of feature data by comparing the obtained plurality of feature data with each other. In claim 2,
A personal authentication method, wherein at the time of authentication, an authentication process is stopped when feature data having a pupil opening index that is close to the obtained pupil opening index by a predetermined degree is not registered for the registrant. In claim 5,
When you stop the authentication process,
Pupil opening index determined at the time of authentication, based on the pupil opening index according to the registered feature data, to estimate the preferred imaging conditions of the iris image,
A personal authentication method characterized by prompting the subject to reacquire an iris image under the estimated photographing conditions. In claim 1,
At the time of registration,
For the registrant, obtain a plurality of iris images having different pupil openings,
From a plurality of feature data and a pupil opening index obtained from the plurality of obtained iris images, a relational expression between the feature data and the pupil opening index is obtained,
Registering a parameter for expressing this relational expression in the iris database in association with the registrant,
At the time of authentication,
A relational expression is obtained from parameters registered in association with a registrant in the iris database, and the pupil opening index obtained at the time of authentication is substituted into the relational expression to obtain the characteristic data of the matching target. And personal authentication method. In claim 7,
At the time of registration, the parameters are thinned out and registered,
A personal authentication method characterized in that at the time of authentication, thinned parameters are restored by interpolation. In claim 1,
At the time of registration,
For the registrant, obtain a plurality of iris images having different pupil openings,
A conversion rule for specifying feature data to be registered from a plurality of feature data obtained from the plurality of acquired iris images, and performing a conversion from the registered feature data to another feature data having a different pupil opening index. ,
The registered feature data and the conversion rule are registered in the iris database in association with the registrant,
At the time of authentication,
An individual using the pupil opening index obtained at the time of authentication based on feature data and a conversion rule registered in association with a registrant in the iris database, wherein the feature data to be collated is generated. Authentication method. In claim 1,
The personal authentication method, wherein the pupil opening index is a ratio between a pupil diameter and an iris diameter in an iris image. A personal authentication method using an iris image,
A first step of obtaining an iris image for the subject;
A second step of obtaining feature data and a pupil opening index from the iris image obtained in the first step;
By referring to the data registered for the registrant in the iris database in which the data registration using the pupil opening index has been performed, the characteristic data to be collated is obtained using the pupil opening index obtained in the second step. The third step;
By comparing the feature data to be checked obtained in the third step with the feature data obtained in the second step, it is determined whether or not the authenticated person is the same as the registrant. A personal authentication method, comprising: In claim 11,
In the iris database, for each registrant, at least one feature data of an iris image is registered together with a pupil opening index,
In the third step, from among at least one feature data registered in the iris database in association with a registrant, a pupil opening index registered along with the feature data and the pupil opening index in the second step A personal authentication method comprising identifying the characteristic data to be matched by comparing the obtained pupil opening index with the obtained pupil opening index. In claim 11,
In the iris database, for each registrant, a parameter representing a relational expression between the feature data of the iris image and the pupil opening index is registered,
In the third step, a relational expression is obtained from the parameters registered in the iris database in association with the registrant, and the pupil opening index obtained in the second step is substituted into the relational expression. And obtaining the characteristic data to be collated. In claim 11,
In the iris database, for each registrant, feature data of an iris image and a conversion rule for performing conversion from the feature data to another feature data having a different pupil opening index are registered,
In the third step, based on the feature data and the conversion rule registered in the iris database in association with the registrant, using the pupil opening index obtained in the second step, A personal authentication method characterized by generating characteristic data. An apparatus for registering data for iris authentication,
Means for obtaining an iris image for the registrant;
Means for calculating feature data and a pupil opening degree index from the iris image;
Means for registering data on the registered person in the iris database using the characteristic data and the pupil opening index. An apparatus for performing personal authentication using an iris image,
Means for acquiring an iris image for the subject,
Means for calculating feature data and a pupil opening degree index from the iris image;
Using the pupil opening index, referring to data registered in association with the registrant in the iris database in which data registration using the pupil opening index is performed, means for obtaining feature data to be collated,
An iris authentication device, comprising: means for comparing the feature data to be compared with the feature data to determine whether or not the person to be authenticated is the same as the registrant. . A program for causing a computer to execute personal authentication using an iris image,
Calculating feature data and a pupil opening degree index from the iris image obtained for the subject;
Using the pupil opening index, referring to data registered for the registrant in the iris database in which data registration using the pupil opening index was performed, and obtaining feature data to be compared;
Determining whether the authenticated person is the same as the registrant by comparing the characteristic data to be compared with the characteristic data.

Images