JP2006326326A - Organism eye judgment method, and organism eye judgment device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To discriminate an organism eye from a photograph eye or an artificial eye with a simple system. <P>SOLUTION: An eye EY of a person to be shot is illuminated coaxially with the optical axis of a camera 11, and then, is shot. At this time, a retinal reflex is caused and the luminance in the pupil region becomes high if the eye EY is a living eye. A living eye judgment section 17 judges whether or not an eye included in the image is a living eye is performed based on the luminance in the pupil region of the image captured by the camera 11. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to so-called biological eye determination that determines whether or not an eye shown in an image is a biological eye, for example, in order to prevent erroneous authentication of a non-biological eye in iris authentication. .

  In recent years, technology for personal authentication using iris images has begun to be used for entrance / exit management of important facilities, ATM (Automated Teller Machine) such as banks, and PC login. In particular, the method described in Patent Document 1 has already been commercialized in various countries around the world, and is becoming a de facto global standard system.

  In the method of Patent Document 1, an iris region is cut out from an image obtained by photographing an iris, and after the iris region is expressed in polar coordinates, 2D Gabor Wavelet filtering is performed to generate an iris code. Then, personal authentication is performed by comparing the generated iris code with the previously registered iris code.

Further, in personal authentication using an iris image, there is a possibility that authentication of an iris photograph or artificial eye is performed erroneously. Techniques for preventing such erroneous authentication are disclosed in Patent Documents 2 and 3. In Patent Document 2, blinking of eyes is detected, or changes in pupil diameter are detected. In Patent Document 3, a change in pupil diameter is observed by applying visible light stimulation.
Japanese Patent No. 3307936 JP 2000-105830 A Japanese Patent No. 3315648

  However, the conventional techniques have the following problems.

  First, Patent Document 2 has a problem that a high-speed shutter is necessary and processing time is required. Also in Patent Document 3, a considerable amount of time is required for processing in order to see changes in pupil diameter.

  In view of the above problems, an object of the present invention is to make it possible to discriminate between the eyes of a living body and photographs and artificial eyes by a simple method.

  In the present invention, a phenomenon called retinal reflection is used. The retinal reflex will be briefly described.

  FIG. 1 is a cross-sectional view of an eyeball and shows a mechanism of retinal reflection. As shown in FIG. 1, in the case of retinal reflection, light is incident through the pupil (a region not covered by the iris) and reflected by the retina, and the reflected light is reversed in the same direction as the incident light. Through the eyeball. For this reason, the entire pupil appears bright in the eye image.

  FIG. 2A is a schematic diagram of an eye image when retinal reflection occurs, and FIG. 2B is a graph showing a luminance profile in the horizontal line L of the image of FIG. It can be seen from FIG. 2B that the luminance is high in the pupil region.

  Such retinal reflection is a phenomenon peculiar to the eyes of a living body and does not occur in a printer image or an artificial eye. The present invention focuses on this point.

  That is, in order to solve the above-described problem, in the present invention, a subject is illuminated coaxially with the optical axis of the camera if it is a living eye, using an image taken by illuminating the optical axis of the camera coaxially. When the retinal reflection occurs, it is determined whether or not the eye shown in the image is a living eye based on the luminance value of the pupil region of the eye shown in the image.

  According to the present invention, in an image captured by illuminating coaxially with the optical axis of the camera, the luminance of the pupil region of the living body's eyes increases due to retina reflection. On the other hand, retinal reflection does not occur in photographs, printer images, or artificial eyes, and the luminance of the pupil region does not become as high as that of the eyes of a living body. Therefore, based on the luminance value of the pupil region of the image, it can be easily determined whether or not the eye shown in the image is a biological eye.

  In the present invention, the first image obtained by illuminating the subject with the optical axis of the camera coaxially and the second image obtained by illuminating with the optical axis different from the optical axis of the camera are used. Based on the luminance value of the pupil region of the eye shown in the first and second images, utilizing the fact that retinal reflection occurs when illuminated in the same axis as the optical axis of the camera if it is a living eye, It is determined whether or not the eye included in the image is a living eye.

  According to the present invention, with respect to the eyes of the living body, in the first image taken by illuminating coaxially with the optical axis of the camera, the luminance of the pupil region increases due to retinal reflection and the light is different from the optical axis of the camera. In the second image captured with the axis illuminated, the luminance of the pupil region is small. That is, in the first and second images, a large luminance difference occurs in the pupil region. On the other hand, with a photograph, a printer image, or a prosthetic eye, the first and second images do not differ as much in brightness as the eyes of the living body in the pupil region. Therefore, based on the luminance value of the pupil region of the image, it can be easily determined whether or not the eye shown in the image is a biological eye. Further, even when a photograph or a printer image when retinal reflection occurs is used, no difference in brightness occurs between the first and second images, so that accurate discrimination can be performed.

  According to the present invention, it is possible to easily determine whether or not the eye shown in the image is a living eye based on the luminance value of the pupil region of the image.

  According to the first aspect of the present invention, as a biological eye determination method, a first step of acquiring an image obtained by illuminating a subject coaxially with the optical axis of a camera, and the eye shown in the image is a living body Based on the luminance value of the image in the pupil region of the eye, whether the eye is a living eye or not, using the fact that retinal reflection occurs when the eye is illuminated coaxially with the optical axis of the camera. And a second step of determining.

  According to the second aspect of the present invention, in the second step, when the difference or ratio between the luminance values of the pupil region and the iris region is larger than a predetermined threshold, the eye is determined to be a living eye. A living eye determination method according to one aspect is provided.

  According to the third aspect of the present invention, in the first step, a plurality of the temporally continuous images are acquired, and in the second step, a predetermined value related to the luminance value in the pupil region obtained from each image is obtained. The living eye determination method according to the first aspect for determining whether or not the eye is a living eye based on the time change of the index, based on the fact that the pupil is shaken if the eye is a living eye. provide.

  According to a fourth aspect of the present invention, there is provided the biological eye determination method according to the third aspect, wherein the predetermined index is an average value of luminance values in a pupil region.

  According to a fifth aspect of the present invention, there is provided the biological eye determination method according to the third aspect, wherein the predetermined index is a ratio of luminance values of a pupil region and an iris region.

  According to a sixth aspect of the present invention, there is provided the biological eye determination method according to the third aspect, wherein the predetermined index is a sum of luminance values of each pixel in the pupil region.

  According to a seventh aspect of the present invention, there is provided the living eye judging method according to the sixth aspect, wherein a value normalized by the area of the iris region is used as the sum of the luminance values.

  According to the eighth aspect of the present invention, as a biological eye determination method, a first image obtained by illuminating a subject coaxially with the optical axis of the camera and illuminating the subject with an optical axis different from the optical axis of the camera are used. The first step of acquiring the second image taken in the above, and whether or not the eyes shown in the first and second images are the eyes of a living body. And a second step of determining based on the luminance value in the pupil region of the eye of the first and second images using the fact that retinal reflection occurs when illuminated coaxially with provide.

  According to the ninth aspect of the present invention, in the second step, a first luminance difference that is a difference in luminance value between a pupil region and an iris region in the first image, and a pupil region in the second image And a second luminance difference that is a difference in luminance value between the iris region and the iris region, and when the absolute value of the difference between the first luminance difference and the second luminance difference is greater than a predetermined threshold, An eighth aspect of the biological eye determination method for determining that the eye is a biological eye is provided.

  According to the tenth aspect of the present invention, in the second step, the first luminance ratio that is the ratio of the luminance values of the pupil region and the iris region in the first image, and the pupil region in the second image A second luminance ratio that is a ratio of luminance values of the iris region to the iris region, and when the ratio between the first luminance ratio and the second luminance ratio is greater than a predetermined threshold, the eye is a living body. An eighth aspect of the method for determining a living eye according to the eighth aspect is provided.

  According to the eleventh aspect of the present invention, as the biological eye determination device, a camera that shoots a subject, an illuminating unit that illuminates the subject coaxially with the optical axis of the camera, and photographic by the camera during illumination by the illuminating unit Taking advantage of the fact that retinal reflection occurs when illuminated in the same axis as the optical axis of the camera if it is a biological eye, whether the eye reflected in the image is a biological eye, An image is provided that includes a living eye determination unit that determines the image based on the luminance value in the pupil region of the eye.

  According to a twelfth aspect of the present invention, a camera for photographing a subject, first illumination means for illuminating the subject coaxially with the optical axis of the camera, and illuminating the subject with an axis different from the optical axis of the camera Second illumination means, a first image taken by the camera during illumination by the first illumination means, and a second image taken by the camera during illumination by the second illumination means In response to whether the eyes shown in the first and second images are the eyes of a living body, if the eyes of the living body are illuminated coaxially with the optical axis of the camera, retinal reflection occurs. Thus, there is provided a living eye determining unit that determines the first and second images based on the luminance value in the pupil region of the eye.

  Embodiments of the present invention will be described below with reference to the drawings.

(First embodiment)
FIG. 4 is a flowchart showing the iris authentication method according to the first embodiment of the present invention. The flow of FIG. 4 includes steps S14 and S15 for performing living eye determination.

  FIG. 5 is an example of the configuration of the iris authentication apparatus according to the present embodiment. In FIG. 5, the photographing apparatus 10 includes a camera 11 for photographing a subject's eye EY (subject), an illumination 12, and a half mirror 13. The light emitted from the illumination 12 is reflected by the half mirror 13 to illuminate the eyes EY. The camera 11 takes an image of the eye EY through the half mirror 13 when illuminated by the illumination 12. At this time, the optical axis of the camera 11 and the illumination 12 are coaxial (coaxial epi-illumination). That is, the illumination means 19 is constituted by the illumination 12 and the half mirror 13. The image processing unit 15 includes an iris authentication unit 16 and a biological eye determination unit 17, and performs iris authentication including biological eye determination on the eye image obtained by the imaging apparatus 10. The imaging device 10 and the biological eye determination unit 17 constitute a biological eye determination device.

  First, in step S <b> 11, an iris image is photographed by the photographing device 10. As shown in FIG. 6, the eye is composed of a pupil A1, an iris A2, and a sclera A3 when viewed from the front. The boundary between the pupil A1 and the iris A2 is referred to as the pupil outer edge E1, and the boundary between the iris A2 and the sclera A3 is referred to as the iris outer edge E2. In step S11, photographing is performed so that at least the iris A2 is included in the image.

  The illumination 12 of the imaging device 10 is composed of, for example, one LED or several LEDs having a near-infrared wavelength region. An advantage of illuminating the eye EY using near infrared light is that the subject does not feel dazzling.

  Further, a visible light cut filter 14 is provided in front of the camera 11 in order to use the captured image for iris authentication. Thereby, the camera 11 receives only the near-infrared component.

  In this embodiment, an LED having a near-infrared wavelength range is used as the light source, but other light sources or other wavelength ranges (for example, visible light) may be used.

  At the time of shooting, only when the camera 11 is exposed, the illumination 12 is synchronized to emit light. Thereby, the burden on the person to be photographed due to the illumination of the eyes can be reduced.

  Next, in step S12, iris authentication is performed using the image photographed in step S11. This process is executed by the iris authentication unit 16. Any method may be used as the iris authentication method, but here, the method described in Patent Document 1 is used. The outline of the method is as follows.

  (1) The pupil outer edge E1 and the iris outer edge E2 are determined, and the iris region is cut out.

  (2) The cut out iris region is converted from the xy orthogonal coordinate system to the rθ polar coordinate system.

  (3) The analysis band is determined (the radial direction is divided into eight rings).

  (4) Applying a multi-scale 2-d Gabor filter and binarizing the signal after the Gabor filter output is an iris code.

  (5) The registered iris code registered in advance and the iris code at the time of authentication are compared (exclusive OR) to calculate the Hamming distance between the two codes.

  (6) If the hamming distance is less than or equal to the threshold, accept as the person, otherwise reject as the other person.

  In step S13, it is checked whether authentication is successful. If accepted as the principal in (6) above, the authentication is successful (Yes), and the process proceeds to step S14. When that is not right (No), it rejects as others and complete | finishes a process.

  In step S14, living eye determination is performed. This process is executed by the living eye determination unit 17, and details will be described later. And when it determines with it not being a living body's eye as a result of determination (it is No at S15), it rejects and complete | finishes a process. On the other hand, when it is determined that the eye is a living body (Yes in S15), the processing ends as authentication success.

  In the flow of FIG. 4, the iris authentication is performed first, and the living eye determination is performed only when the authentication is successful. However, as shown in FIG. 7, the living eye determination is performed first (S22). Only when it is determined to be an eye (Yes in S23), iris authentication may be performed (S24). In any method, in order to be finally authenticated, it is necessary not to be rejected in both the biological eye determination and the iris authentication.

  Here, the biological eye determination in step S14 will be described in detail according to the flow of FIG. The processing here corresponds to the living eye determination method according to the present embodiment.

  First, in step S31, the pupil region is determined. Here, as shown in the flow of FIG. 4, when the iris authentication S12 is performed prior to the biological eye determination S14, the position of the pupil has already been obtained in the iris authentication process. Good. On the other hand, when the living eye determination S22 is performed prior to the iris authentication S24 as in the flow of FIG. 7, the pupil region is determined here. Any method can be used to determine the pupil region. For example, the method described in Patent Document 1 may be used.

  Next, in step S32, it is determined based on the luminance value of the pupil region whether or not the eye shown in the image is a living eye. When photographing using coaxial epi-illumination as in the present embodiment, the brightness of the entire pupil region of the image is increased by retinal reflection as shown in FIG. On the other hand, retinal reflection does not occur in eye photographs or artificial eyes even with coaxial epi-illumination. In this embodiment, this point is used to discriminate between the eyes of a living body and photographs or artificial eyes.

  Here, an average value of luminance values in the pupil region is used for the living eye determination. When the average value is larger than the predetermined threshold value TH1, it is determined that retinal reflection is occurring, and it is determined that the eye shown in the image is a living eye.

  In addition, in order to avoid the influence of corneal reflection in the pupil region, when calculating the average value of the luminance values in the pupil region, only pixels whose luminance value is smaller than the predetermined threshold value TH2 may be used. FIG. 3 is a diagram showing the mechanism of corneal reflection. As shown in FIG. 3, in the case of corneal reflection, light is reflected by the cornea on the surface of the eyeball.

  One of the features of corneal reflection is that the luminance is higher than that of the retinal reflection and the luminance is likely to be saturated. For this reason, the influence of corneal reflection can be easily excluded by appropriately setting the threshold value TH2.

  FIG. 9 is a schematic diagram of an image obtained by photographing an actual eye using coaxial epi-illumination, and FIG. 10 is a schematic diagram of a living eye image obtained by photographing the printer output of an iris image using coaxial epi-illumination. In the image obtained by photographing the actual eye, retinal reflection occurs as shown in FIG. 9, and in the image obtained by photographing the printer output, retinal reflection does not occur as shown in FIG. It is assumed that the average value of the luminance values of the pupil region (excluding the luminance saturation region) is P1 in FIG. 9 and P2 in FIG. Therefore, by setting the threshold TH1 as P2 <TH1 <P1, it is possible to determine whether or not the eye is a living body.

  Of course, the brightness of the image changes depending on the intensity of illumination, the aperture of the lens, the black level and gain of the camera, and the value of the threshold value TH1 may be adjusted according to those values.

  In addition, as can be seen from FIG. 2B, in the case of retinal reflection, the luminance value in the pupil region increases almost uniformly, so whether the variance or standard deviation of the luminance value in the pupil region is smaller than a predetermined threshold value TH3. No may be added to the determination condition of the eyes of the living body. In FIG. 9, when the standard deviation of the luminance value of the pupil region (excluding the luminance saturation region) is σ1, the threshold value TH3 may be set such that TH3> σ1. By determining not only the average luminance value of the pupil region but also the standard deviation, more accurate living eye determination can be performed.

  Further, in each image, the luminance values of the pupil region and the iris region may be compared. For example, when the luminance value of the pupil region is larger than the luminance value of the iris region, it is determined that the eye is a living body. In the image of FIG. 10, it is determined that the eye is not a living eye by utilizing the fact that the luminance value P2 of the pupil region is smaller than the luminance value I2 of the iris region. On the other hand, in the image of FIG. 9, it is determined that the eye is a living body by utilizing the fact that the luminance value P1 of the pupil region is larger than the luminance value I1 of the iris region.

  Further, whether or not the difference between the luminance values of the pupil region and the iris region is larger than a predetermined threshold value TH4 may be used as a determination condition for the eyes of the living body. The difference in luminance value is P1-I1 in the image of FIG. 9 and P2-I2 in the image of FIG. Therefore, by setting TH4 as P2-I2 <TH4 <P1-I1, FIG. 10 can be determined as an image that is not the eyes of a living body.

  Or it is good also considering whether the ratio of the luminance value of a pupil area | region and an iris area | region is larger than predetermined | prescribed threshold value TH5, as a determination condition of a biological eye. The ratio of the luminance values is P1 / I1 in the image of FIG. 9, and P2 / I2 in the image of FIG. Therefore, by setting TH5 as P2 / I2 <TH5 <P1 / I1, it can be determined that FIG. 10 is not a biological eye.

  As described above, according to the present embodiment, it is possible to distinguish a living eye from a photograph or a prosthetic eye based on the luminance value of the pupil region from an image obtained by illuminating the subject coaxially with the optical axis of the camera. Can do. Therefore, it is possible to realize the biological eye determination by a simple method. For example, in personal authentication, it is possible to prevent erroneous authentication of an object that is not a biological eye, and to improve reliability.

  In order to obtain good retinal reflection, the eye EY of the subject needs to face the camera 11. For this purpose, a marker for guiding the line of sight may be provided at the center of the transparent cover provided in front of the visible light cut filter 14 or the photographing apparatus 10. Further, a visible light illumination for visual line guidance may be separately provided next to the illumination 12. In this case, the visible light illumination is preferably set to a minimum brightness that can be visually recognized so that the subject is not dazzled.

  In the flow of FIG. 4, it is assumed that the biometric eye determination is performed after the iris authentication, and in the flow of FIG. 7, the iris authentication is performed after the biometric eye determination. However, the iris authentication and the biometric eye determination are not necessarily performed in series. There is no need to do this. For example, two arithmetic devices such as the iris authentication unit 16 and the biological eye determination unit 17 may execute the processing in parallel.

<About "coaxial">
In the present embodiment, the subject is illuminated coaxially with the optical axis of the camera. In this case, “coaxial” does not mean a strict coaxial, but “almost coaxial” to the extent that retina reflection can be photographed. " That is, the “coaxial” in the present invention includes not only a strict coaxial, but also a schematic coaxial within a range where retinal reflection can be photographed.

  With reference to FIG. 11 and FIG. 12, the range in which retinal reflection can be imaged will be described. Incident light of illumination is refracted by a crystalline lens corresponding to a lens and forms an image on the retina. The illumination light imaged on the retina is refracted by the lens and goes out of the eyeball. When the human eye focuses on the illumination and there is no aberration of the crystalline lens corresponding to the lens, the illumination light imaged on the retina is condensed again at the illumination position as shown in FIG. To do. Therefore, when the camera is placed in the shaded area in FIG. 11, retinal reflection can be photographed.

  On the other hand, when the human eye is not focused on the illumination or there is an aberration of the crystalline lens, the region where the retina reflection can be photographed becomes wider as shown in FIG.

  Accordingly, when the optical axis of the camera is included in the hatched area in FIG. 11 or FIG. 12, it is defined as being substantially coaxial so that retina reflection can be photographed, and included in the “coaxial” of the present invention.

  Incidentally, an example of retinal reflection that occurs even when the illumination and the camera optical axis are misaligned is the red-eye phenomenon caused by the camera flash. In a normal camera, the lens position and the flash position are different, but since the distance is smaller than the shooting distance, red eyes, which are retinal reflections, are shot although they are not perfectly coaxial.

<Other configuration of photographing apparatus>
In this embodiment, the half mirror 13 is provided in order to make the optical axis of the illumination 12 and the camera 11 coaxial, but the positions of the illumination 12 and the camera 11 may be set opposite to the half mirror 13. In this case, the light of the illumination 12 passes through the half mirror 13 to illuminate the eyes EY, and the camera 11 captures an image reflected by the half mirror 13.

  Further, the configuration in which the illumination and the optical axis of the camera are coaxial can be realized without using a half mirror. For example, in the photographing apparatus 10A shown in FIG. 13, an illumination 12A as an illumination unit is disposed on the optical axis in front of the camera 11. However, in this configuration, it is necessary to provide the light shielding means 18 between the illumination 12A and the camera 11 so that the light of the illumination 12A does not directly enter the camera 11. Naturally, since the subject cannot be photographed through the area of the light shielding means 18, the image photographed by the photographing apparatus 10A in FIG. 13 includes the light shielding area CR in the pupil area A1, as shown in FIG. . However, the iris authentication is not a problem because it is only necessary to obtain an image of the iris region A2.

  Further, in the photographing apparatus 10B of FIG. 15, the illumination 12B as the illumination means is arranged slightly shifted from the optical axis of the camera 11. However, the position of the camera 11 with respect to the illumination 12B is included in an area where retinal reflection can be photographed as shown in FIG. That is, the eye EY of the subject is illuminated so-called “coaxial” with the optical axis of the camera 11. Further, as a modification of FIG. 15, a plurality of illuminations 12c as illumination means may be arranged around the camera lens 11a as shown in FIG.

(Second Embodiment)
The iris authentication method according to the second embodiment of the present invention has the same basic processing flow as that of the first embodiment. The difference is that a second image obtained by illuminating the subject with an optical axis different from the optical axis of the camera is obtained separately from the first image obtained by illuminating the subject coaxially with the optical axis of the camera. This is the point of performing living eye determination using the first and second images.

  FIG. 17 shows an example of the configuration of the iris authentication apparatus according to the present embodiment, and the same reference numerals as those in FIG. The imaging device 20 in FIG. 17 is different from the imaging device 10 in FIG. 5 in that an illumination 21 is provided in addition to the illumination 12. The illumination 21 directly illuminates the subject's eye EY with an axis different from the optical axis of the camera 11 (that is, non-coaxial illumination that is not coaxial epi-illumination). The illumination 12 and the half mirror 13 constitute a first illumination means 19, and the illumination 21 constitutes a second illumination means. The image processing unit 22 includes an iris authentication unit 23 and a biological eye determination unit 24, and performs iris authentication including biological eye determination on the eye image obtained by the imaging device 20. The imaging device 20 and the biological eye determination unit 24 constitute a biological eye determination device.

  The processing in this embodiment will be described here with reference to the flow of FIG. Of course, as shown in FIG. 7, the biological eye determination may be performed before the iris authentication, or the iris authentication and the biological eye determination may be performed in parallel.

  First, in step S <b> 11, an iris image is photographed by the photographing device 20. Here, photographing is performed for each of the case where only the illumination 12 is caused to emit light and the case where only the illumination 21 is caused to emit light. The image when only the illumination 12 is emitted, that is, the image when the coaxial epi-illumination is used is the first image, and the image when only the illumination 21 is emitted, that is, when the image is not the coaxial epi-illumination, is the second image.

  Next, in step S12, iris authentication is performed using the image photographed in step S11. This process is executed by the iris authentication unit 23. For iris authentication, at least one of the first and second images is used. Here, authentication is performed using the second image. Of course, authentication may be performed using the first image. In addition, authentication is performed using both the first and second images. For example, if authentication succeeds in either of them, the authentication is finally successful, and if authentication is successful in both, the authentication is finally successful. The authentication policy may be determined according to the security level. The iris authentication method is as shown in the first embodiment.

  In step S13, it is checked whether authentication is successful. If the authentication is successful (Yes), the process proceeds to step S14. If not (No), it is rejected as another person and the process is terminated.

  In step S14, living eye determination is performed. Details will be described later. And when it determines with it not being a living body's eye as a result of determination (it is No at S15), it rejects and complete | finishes a process. On the other hand, when it is determined that the eye is a living body (Yes in S15), the processing ends as authentication success.

  Here, the biological eye determination in step S14 will be described in detail according to the flow of FIG. The processing here corresponds to the living eye determination method according to the present embodiment.

  First, in step S31, the pupil region is determined for the first and second images. Here, for the second image, the position of the pupil has already been obtained in step S12, so that the pupil region may be determined for the first image. However, when the shooting interval of the first and second images is short, for example, when shooting continuously with a camera of 30 frames / s, the pupil region in the first and second images is almost changed. It is not considered. Therefore, the pupil position in the second image may be adopted as the pupil position in the first image. Further, when the living eye determination S22 is performed prior to the iris authentication S24 as in the flow of FIG. 7, the pupil region is determined here. Any method can be used to determine the pupil region. For example, the method described in Patent Document 1 may be used.

  Next, in step S32, based on the luminance value of the pupil region in the first and second images, it is determined whether or not the eye shown in the image is a living eye. In the case of a living eye, in the case of coaxial epi-illumination, the luminance of the entire pupil region increases due to retinal reflection, while in the case of non-coaxial epi-illumination, the luminance of the pupil region decreases. In the present embodiment, this point is used to distinguish a living eye from a photograph or an artificial eye.

  Here, an average value of luminance values in the pupil region is used for the living eye determination. In the first and second images, assuming that the average values of the luminance values of the pupil region are Pa and Pb, respectively, it is determined that retinal reflection occurs when the luminance difference (Pa−Pb) is larger than a predetermined threshold value TH1. Then, it is determined that the eye shown in the image is a biological eye.

  18 and 19 are examples of a second image taken using non-coaxial illumination. FIG. 18 is a photograph of actual eyes, and FIG. 19 is a photograph of printer output. Also, assuming that FIGS. 9 and 10 are obtained as the first image taken using the coaxial incident illumination, the luminance difference (Pa−Pb) is P1− in the actual eye image (FIGS. 9 and 18). P3, and P2-P4 in the images (FIGS. 10 and 19) obtained by photographing the printer output. Therefore, by setting the threshold value TH1 as P2-P4 <TH1 <P1-P3, it is possible to discriminate between eyes of a living body and eyes that are not a living body.

  When the average luminance value Pa of the pupil region in the first image is larger than the predetermined threshold value TH2, and the average luminance value Pb of the pupil region in the second image is smaller than the predetermined threshold value TH3, May be determined.

  Further, a difference in luminance value between the pupil region and the iris region may be used. If the average luminance value of the iris region in the first image is Ia and the average luminance value of the iris region in the second image is Ib, for example, the luminance difference in the first image, that is, the first difference (Pa-Ia) When the absolute value of the brightness difference in the second image, that is, the difference between the second differences (Pb−Ib) is larger than the threshold value TH4, it is determined that the eye is a living body. In the case of the previous image example, for actual eyes, the luminance difference (Pa-Ia) in the first image (FIG. 9) is P1-I1, and the luminance difference (Pb-Ib) in the second image (FIG. 18). Is P3-I3, and the absolute value of the difference is | (P1-I1)-(P3-I3) |). On the other hand, regarding the printer output, the luminance difference (Pa-Ia) in the first image (FIG. 10) is P2-I2, and the luminance difference (Pb-Ib) in the second image (FIG. 19) is P4-I4. The absolute value of the difference is | (P2-I2)-(P4-I4) |. Therefore, by setting the threshold TH4 as | (P2-I2) − (P4-I4) | <TH4 <| (P1-I1) − (P3-I3) | Can be determined. This calculates the difference between the first image and the second image for the luminance value of the pupil region and the difference between the first image and the second image for the luminance value of the iris region, and calculates the absolute value of the difference. Is equivalent to using. That is, | (Pa-Ia)-(Pb-Ib) | = | (Pa-Pb)-(Ia-Ib) |.

  When the luminance difference (Pa-Ia) in the first image is larger than the predetermined threshold TH6 and the luminance difference (Pb-Ib) in the second image is smaller than the predetermined threshold TH5, It may be determined that

  Or you may use the ratio of the luminance value of a pupil area | region and an iris area | region. For example, when the luminance ratio in the first image, that is, the first ratio Pa / Ia, and the luminance ratio in the second image, that is, the ratio of the second ratio Pb / Ib is larger than the threshold value TH7, it is a living eye. Is determined. In the case of the previous image example, for the actual eyes, the luminance ratio Pa / Ia in the first image (FIG. 9) is P1 / I1, and the luminance ratio Pb / Ib in the second image (FIG. 18) is P3 / I3. The ratio is P1 / I1 / (P3 / I3). On the other hand, for the printer output, the luminance ratio Pa / Ia in the first image (FIG. 10) is P2 / I2, and the luminance ratio Pb / Ib in the second image (FIG. 19) is P4 / I4. P2 / I2 / (P4 / I4). Therefore, by setting the threshold value TH7 as P2 / I2 / (P4 / I4) <TH7 <P1 / I1 / (P3 / I3), it is possible to determine whether the eye is a living body. This calculates the ratio of the first image to the second image for the luminance value of the pupil region and the ratio of the first image to the second image for the luminance value of the iris region, and uses those ratios. Is equivalent to That is, (Pa / Ia) / (Pb / Ib) = (Pa / Pb) / (Ia / Ib).

  Further, when the luminance ratio Pa / Ia in the first image is larger than the predetermined threshold TH9 and the luminance ratio Pb / Ib in the second image is smaller than the predetermined threshold TH8, it is determined that the eye is a living body. May be.

  In addition, in order to avoid the influence of corneal reflection in the pupil region, when calculating the average value of the luminance values in the pupil region, only pixels whose luminance value is smaller than the predetermined threshold value TH10 may be used. Further, in the case of retinal reflection, the luminance value in the pupil region increases substantially uniformly. Therefore, whether or not the variance or standard deviation of the luminance value in the pupil region is smaller than a predetermined threshold TH11 is determined. It may be added to the determination condition.

  As described above, according to the present embodiment, the luminance of the pupil region of both images using an image obtained by illuminating the subject coaxially with the optical axis of the camera and an image illuminated with an axis different from the optical axis of the camera. Based on the value, it is possible to distinguish a living eye from a photograph or an artificial eye. Therefore, accurate biological eye determination can be realized by a simple method. For example, in personal authentication, it is possible to prevent erroneous authentication of non-biological eyes, and to improve reliability.

  In the present embodiment, only the illumination 12 is emitted when the first image is photographed, and only the illumination 21 is emitted when the second image is photographed. However, when the first image is photographed. The illumination 12 and the illumination 21 may be emitted simultaneously. In this case, if the brightness of the iris region is ensured by the illumination 21 and the illumination 12 is set to the minimum light amount that causes retinal reflection, the burden on the eyes due to the light being irradiated near the center of the retina can be reduced. it can.

  When the camera and the illumination are made non-coaxial, they should be shifted from the coax so that retinal reflection does not occur.

(Third embodiment)
The iris authentication method according to the third embodiment of the present invention has the same basic processing flow as that of the first embodiment. The difference is that when a subject is illuminated coaxially with the optical axis of the camera and an image is taken, a plurality of images are acquired, and a living eye determination is performed using these images.

  As an iris authentication apparatus according to this embodiment, the apparatus shown in FIG. 5 is used as in the first embodiment. The constituent elements in FIG. 5 are the same as those described in the first embodiment, and a description thereof will be omitted.

  The processing in this embodiment will be described here with reference to the flow of FIG. Of course, as shown in FIG. 7, the biological eye determination may be performed before the iris authentication, or the iris authentication and the biological eye determination may be performed in parallel.

  First, in step S11, a plurality of iris images are photographed by the photographing device 10. Here, an iris image is shot at a shooting speed of 30 frames per second, and the illumination 12 is emitted in synchronization with the exposure of the camera 11.

  Next, in step S12, iris authentication is performed using a plurality of images taken in step S11. This process is executed by the iris authentication unit 16.

  Here, it is assumed that image quality determination is performed separately and one image having the maximum evaluation value for image quality determination is used for iris authentication. As the image quality, a focus value, a degree of opening of the eyelid, and the like can be used. Of course, any image among a plurality of photographed iris images may be used for authentication. In addition, authentication is performed using all iris images, and for example, if authentication is successful in any of them, the authentication is finally successful, and if authentication is successful with a predetermined ratio of iris images, the authentication may be successful. The authentication policy may be determined according to the security level. The iris authentication method is as shown in the first embodiment.

  In step S13, it is checked whether authentication is successful. If the authentication is successful (Yes), the process proceeds to step S14. If not (No), it is rejected as another person and the process is terminated.

  In step S14, living eye determination is performed. This process is executed by the living eye determination unit 17. Details will be described later. And when it determines with it not being a living body's eye as a result of determination (it is No at S15), it rejects and complete | finishes a process. On the other hand, when it is determined that the eye is a living body (Yes in S15), the processing ends as authentication success.

  Here, the living eye determination in step S14 in the present embodiment will be described in detail according to the flow of FIG. The processing here corresponds to the living eye determination method according to the present embodiment.

  First, in step S31, the pupil region is determined for a plurality of shot iris images. When continuously shooting at a shooting speed of 30 frames per second, it is considered that the pupil area in the plurality of iris images has hardly changed, so only the first image is searched with the search area for the pupil area enlarged, For the second and subsequent frames, the amount of calculation can be reduced by searching only the periphery of the pupil region in the previous frame as an initial search value. Any method can be used to determine the pupil region. For example, the method described in Patent Document 1 may be used.

  Next, in step S32, it is determined based on the luminance value of the pupil region in the plurality of images whether or not the eyes shown in the images are biological eyes. Here, the eyes of the living body repeatedly contract and dilate slightly even when the surrounding brightness is constant. That is, so-called pupil shaking is caused. In this embodiment, the time change of a predetermined index based on the luminance value in the pupil region is obtained from a plurality of images taken under coaxial illumination that would cause retinal reflection if it is a living eye. And the presence or absence similar to a pupil is detected from the time change of this parameter | index, and it is determined whether it is a biological eye.

  Here, the degree of pupil fluctuation, that is, the contraction / expansion of the pupil, can be directly measured from the temporal change of the pupil diameter. However, as a result of actual experiments conducted by the inventors of the present application, it has been found that, for the following reasons, it is difficult to measure changes in pupil diameter with high accuracy in normal iris authentication.

  That is, generally, an iris image used for iris authentication is photographed with a resolution such that the diameter of the iris is about 200 pixels from the camera specifications and the like. In a normal brightness environment, the pupil opening degree is about 0.40 in terms of the pupil diameter / iris diameter ratio, although there are some differences between persons. At this time, the pupil diameter is about 80 pixels. Further, according to the experiments by the inventors of the present application, the fluctuation range of the pupil diameter due to the fluctuation of the pupil under such conditions was 3.7 pixels as a standard deviation (an average value of four subjects). It is expected that it will be considerably difficult to accurately measure a periodic variation of about 3 to 4 pixels with a standard deviation because there is actually a detection error of a pupil region and the like. Therefore, it is considered that it is difficult to detect a change in pupil diameter with sufficient accuracy in a general imaging environment when performing iris authentication.

  Therefore, in the present embodiment, the presence / absence of a minute change in the pupil diameter is determined based on the time change of the index related to the luminance value in the pupil region under the coaxial illumination. When retinal reflection occurs due to coaxial illumination, the brightness of the pupil region increases as the amount of light that passes through the pupil and reaches the retina increases, that is, increases as the pupil diameter increases. Therefore, by using the luminance value of the pupil region (that is, retinal reflection) or the temporal change of the index related to the luminance value, it is possible to determine whether or not the pupil is contracted or dilated.

  Various predetermined indexes relating to the luminance value are conceivable. For example, an average value of luminance values of pixels included in the pupil region may be used as a predetermined index. In this case, when specular reflection (corneal reflection) occurs in the pupil region, an average value of luminance values in the region excluding the corneal reflection region may be calculated. Since the luminance value of the corneal reflection region is usually extremely large, the range can be specified by providing an appropriate threshold value for the luminance value.

  Further, a ratio of luminance values between the pupil region and the iris region may be used as a predetermined index. Thereby, it is possible to provide robustness against changes in the environment.

  As a change in the environment, for example, it is conceivable that the distance between the coaxial illumination and the subject changes during shooting of a plurality of images. In a fixed type imaging device such as an entrance / exit management device, the distance between the coaxial illumination and the subject changes as the subject's head moves. Further, in the handy type photographing apparatus, the distance between the coaxial illumination and the subject changes when the hand holding the device moves or the head of the subject person moves. If the illumination is not perfectly parallel light and the distance to the illumination changes, the luminance value of the pupil region also changes accordingly, so the pupil region alone can accurately measure pupil fluctuation. It may not always be possible.

  On the other hand, the luminance value of the iris region does not change under constant illumination, but when the distance between the coaxial illumination and the subject changes, it changes in accordance with the luminance value of the pupil region. Therefore, by using the ratio of the luminance values of the pupil region and the iris region, it is possible to accurately measure the luminance change due to only the contraction / expansion of the pupil, excluding the influence of the environmental change.

  Further, the sum of the luminance values of the respective pixels in the pupil region may be used as a predetermined index. That is, when the pupil is dilated, the luminance of the pupil region due to retinal reflection increases as described above, and the area of the pupil region also increases. On the other hand, when the pupil contracts, the luminance in the pupil region decreases and the area of the pupil region also decreases. Therefore, by using the sum of the luminance values in the pupil region as an index, the temporal change of the index due to pupil fluctuation is more emphasized than when the average value of luminance values is used. Therefore, S / N with respect to measurement noise can be improved, and measurement accuracy can be improved.

Furthermore, a value normalized by the area of the iris region may be used as the sum of the luminance values of the pixels in the pupil region. For example, when the distance between the illumination and the subject changes, the resolution changes in addition to the luminance of the pupil region. Therefore, the sum of the luminance values in the pupil region is normalized by the luminance value (for example, an average value) of the iris region in order to remove the influence of the luminance change, and further normalized by the area of the iris region. Thereby, the influence of the change in resolution can also be removed. As the area of the iris region, for example, the area of a circle representing the outer edge of the iris (see FIG. 6) (πr ² if the iris radius is r pixels) may be used. Further, the area of the iris region that is actually exposed may be used. The same effect can be obtained by normalizing using the sum of the luminance values of the pixels in the iris region instead of normalizing with the luminance value and area of the iris region. Further, the process of normalizing with the luminance value of the iris region may be omitted, and normalization may be performed only with the area.

  FIG. 20 is a schematic diagram showing the time change of the index related to the luminance value of the pupil region obtained in this way. Although the degree of fluctuation of the vertical axis varies depending on how the index is taken, for a living eye, a periodic fluctuation due to pupil fluctuation as shown in FIG. 20 is obtained.

  Next, it is determined whether or not it is a living eye from the index time change as shown in FIG. Here, the Fourier transform is applied to the time change of the index as shown in FIG. 20, and the determination is performed using the frequency characteristics. FIG. 21 is a graph showing the relationship between the frequency and the power obtained from the time change of the index for the living eye. FIG. 21 uses the luminance sum in the pupil region as a predetermined index, and plots the frequency characteristics of 40 sequences (4 eyes × 10 sequences) averaged. The horizontal axis represents frequency and the vertical axis represents power, both of which are logarithmic. FIG. 21 also shows a broken line obtained by approximating the plot to a straight line. From FIG. 21, it can be seen that the logarithm of power decreases almost linearly with respect to the logarithm of frequency, that is, has a 1 / f characteristic. Therefore, it is possible to perform a living eye determination based on whether or not the frequency characteristic of the sequence photographed at the time of authentication is a 1 / f characteristic.

  In addition, when the imaging resolution is sufficient, or when the change in pupil diameter can be increased (for example, when changing the illuminance of visible light in the imaging environment, however, the illumination intensity of near-infrared light for causing retinal reflection is constant. ) Can accurately obtain changes in pupil diameter. In this case, the biological eye determination may be performed based on the frequency characteristics of the temporal variation of the pupil diameter.

  Moreover, you may use together a pupil diameter and the luminance value of a pupil area | region. For example, the biological eye determination may be performed using whether or not there is a correlation between the pupil diameter and the luminance value of the pupil region. FIG. 22 is a graph showing the relationship between the pupil diameter and the iris luminance / pupil luminance ratio. FIG. 22 shows an iris image taken at 30 frames per second for two persons while changing the visible light intensity in the shooting environment. As can be seen from FIG. 22, as the pupil diameter increases, the pupil luminance increases, and as a result, the iris luminance / pupil luminance ratio decreases. Biological eye determination may be performed based on whether or not there is such a correlation.

  As described above, according to the present embodiment, the time change of a predetermined index related to the luminance value of the pupil region is obtained from a plurality of images obtained by continuously illuminating the subject coaxially with the optical axis of the camera. Ask. Then, based on the time change of the index, it is possible to distinguish between the eyes of a living body and photographs or artificial eyes. Therefore, accurate biological eye determination can be realized by a simple method. For example, in personal authentication, it is possible to prevent erroneous authentication of non-biological eyes, and to improve reliability.

  In the above-described embodiment, an example in which LEDs are used as illumination is described. The LED is a light source that is increasingly used due to the advantages of low power consumption and long life. In the international standard IEC 60825-1 (JIS C6802), a safety standard is defined as a light source that conforms to a laser. Of course, when commercializing, it must be set to meet this safety standard, but when used for coaxial epi-illumination for biological eye determination, even if the safety standard is met, people with weak eyes And for those who are tired, they can feel burdened.

  In preparation for such a case, it is preferable that the illumination is emitted only during the camera exposure time (for example, 16 ms or 33 ms). Alternatively, there may be a limit on the number of times that the living eye determination is executed. For example, the number of authentications per day is counted for each person, and when the number of authentications exceeds the upper limit value, the vital eye determination is not performed. Or you may make it perform other biological eye determination which does not use a coaxial epi-illumination.

  In the above-described embodiments, the biological eye determination is performed together with iris authentication, but may be used for other purposes. In addition, although it was performed on a near-infrared light image, the living eye determination method of the present invention is not limited to an image under near-infrared light illumination, and can of course be applied even under visible light illumination. It is. In this case, the visible light cut filter 14 is not used. The iris authentication image may be taken separately, for example.

  Further, in each of the above-described embodiments, the living eye determination is performed by the image processing unit attached to the imaging apparatus, but the present invention is not limited to such a configuration. For example, an image of the eye is photographed by a portable terminal equipped with the photographing apparatus 10 as shown in FIG. And in a server, you may make it perform a living eye determination by receiving the image. That is, an image of an eye photographed with coaxial epi-illumination is acquired, or an image of an eye photographed with non-co-axial epi-illumination is also acquired, and whether the eye reflected in the image is a living eye A configuration in which processing for determining whether or not is executed is also included in the present invention.

  Since the present invention can easily determine whether or not the eye shown in the image is a living eye based on the luminance value of the pupil region of the image, for example, in personal authentication using an iris image, With this configuration, it is possible to eliminate unauthorized impersonation using photographs or artificial eyes.

It is sectional drawing of an eyeball, and is a figure which shows the mechanism of retinal reflection. (A) is an eye image when retinal reflection occurs, and (b) is a graph showing a luminance distribution in the image of (a). It is a figure which shows the mechanism of corneal reflection. It is a flowchart which shows the iris authentication method including living-eye determination based on each embodiment of this invention. It is an example of the structure of the iris authentication apparatus containing the biological eye determination apparatus based on the 1st Embodiment of this invention. It is a front view which shows the structure of eyes. It is a flowchart which shows the other example of the iris authentication method including living-eye determination based on each embodiment of this invention. It is a flowchart which shows a biological eye determination method. It is a schematic diagram of the image image | photographed with the coaxial epi-illumination. It is a schematic diagram of the image which image | photographed the printer output with coaxial epi-illumination. It is a figure which shows the range which can image | photograph a retinal reflection. It is a figure which shows the range which can image | photograph a retinal reflection. It is another example of a structure of the imaging device which concerns on each embodiment of this invention. It is a figure which shows typically the image image | photographed with the imaging device of FIG. It is another example of a structure of the imaging device which concerns on each embodiment of this invention. It is a modification of the illumination arrangement | positioning of FIG. It is an example of the structure of the iris authentication apparatus containing the biological eye determination apparatus based on the 2nd Embodiment of this invention. It is a schematic diagram of the image image | photographed by the non-coaxial illumination of the biological body. It is a schematic diagram of the image which image | photographed the printer output with non-coaxial illumination. It is a schematic diagram which shows the time change of the predetermined | prescribed parameter | index which concerns on the luminance value of a pupil area | region based on the 3rd Embodiment of this invention. It is an example of the frequency characteristic obtained from the time change of the parameter | index like FIG. It is a graph which shows the relationship between a pupil diameter and iris luminance / pupil luminance ratio.

Explanation of symbols

10, 10A, 10B, 20 Imaging device 11 Camera 12 Illumination 12A, 12B, 12C Illumination (illumination means)
13 Half mirrors 15 and 22 Image processing units 16 and 23 Iris authentication units 17 and 24 Living eye determination unit 19 Illumination means (first illumination means)
21 Illumination (second illumination means

Claims (12)

A first step of acquiring an image obtained by illuminating a subject coaxially with the optical axis of the camera;
Whether or not the eye shown in the image is a living eye, if the eye is a living eye, utilizing the fact that retinal reflection occurs when illuminated coaxially with the optical axis of the camera, the eye of the image And a second step of determining based on the luminance value in the pupil region of the living eye. In claim 1,
In the second step, when the difference or ratio of the luminance values between the pupil region and the iris region is larger than a predetermined threshold, the eye is determined to be a living eye. In claim 1,
In the first step, a plurality of the images that are temporally continuous are acquired,
In the second step, based on the time change of a predetermined index related to the luminance value in the pupil region obtained from each image, using the fact that the pupil is shaken if it is a living eye, A biological eye determination method characterized by determining whether or not an eye is a biological eye. In claim 3,
The biological eye determination method, wherein the predetermined index is an average value of luminance values in a pupil region. In claim 3,
The biological eye determination method, wherein the predetermined index is a ratio of luminance values of a pupil region and an iris region. In claim 3,
The biological eye determination method, wherein the predetermined index is a sum of luminance values of pixels in a pupil region. In claim 6,
A biological eye determination method using a value normalized by an area of an iris region as the sum of the luminance values. A first step of acquiring a first image captured by illuminating the subject coaxially with the optical axis of the camera and a second image captured by illuminating the subject with an optical axis different from the optical axis of the camera; ,
Whether or not the eyes shown in the first and second images are the eyes of a living body, if the eyes of the living body, utilizing the fact that retinal reflection occurs when illuminated coaxially with the optical axis of the camera, A biological eye determination method comprising: a second step of determining based on a luminance value in the pupil region of the eye of the first and second images. In claim 8,
In the second step,
A first luminance difference that is a difference in luminance value between the pupil region and the iris region in the first image, and a second luminance difference that is a difference in luminance value between the pupil region and the iris region in the second image. And
A biological eye determination method, wherein when the absolute value of the difference between the first luminance difference and the second luminance difference is larger than a predetermined threshold, the eye is determined to be a living eye. In claim 8,
In the second step,
A first luminance ratio that is a ratio of luminance values of the pupil region and the iris region in the first image, and a second luminance ratio that is a ratio of luminance values of the pupil region and the iris region in the second image. And
A biological eye determination method, wherein when the ratio between the first luminance ratio and the second luminance ratio is greater than a predetermined threshold, the eye is determined to be a living eye. A camera to shoot the subject,
Illumination means for illuminating the subject coaxially with the optical axis of the camera;
When an image photographed by the camera is received at the time of illumination by the illuminating means, and whether or not the eye shown in the image is a living eye, if it is a living eye, it is illuminated coaxially with the optical axis of the camera A biological eye determination device comprising: a biological eye determination unit that determines based on a luminance value in a pupil region of the eye using the occurrence of retinal reflection. A camera to shoot the subject,
First illumination means for illuminating the subject coaxially with the optical axis of the camera;
Second illumination means for illuminating the subject with an axis different from the optical axis of the camera;
Receiving the first image taken by the camera during illumination by the first illumination means and the second image taken by the camera during illumination by the second illumination means; Whether or not the eye shown in the image is a biological eye, if the biological eye is illuminated with the optical axis of the camera coaxially, the first and second retinal reflections occur. A biological eye determination device, comprising: a biological eye determination unit that determines based on a luminance value in a pupil region of the eye of the image.

Images