JP4296008B2 - Personal authentication system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a personal authentication system, and more particularly to an improvement of a personal authentication system with improved authentication accuracy.
[0002]
[Prior art]
Conventionally, personal authentication has been performed as one method for ensuring security. For example, at the entrance / exit gate portion of the facility, personal authentication for permitting entrance / exit and managing / monitoring the entrance / exit is performed. As a method of personal authentication, a face image of a person to be authenticated (person) is registered in advance, and personal authentication is performed by comparing and collating the face image acquired on the spot and the registered face image at the time of authentication. There is a face recognition system.
[0003]
In this face authentication system, feature points such as the shape and size of eyes, nose, mouth, etc., and the geometric positional relationship between each other are registered from the face image acquired at the time of registration. deep. Then, in the case of actual authentication, the shape and size of the eyes, nose, mouth, etc. obtained from the face image of the person to be authenticated (the person who has requested authentication) acquired in the same way as during registration, and the mutual geometry If a feature point such as a target positional relationship is acquired and compared with a registered feature point, it is determined whether or not it is a registered person. For example, permission to enter and exit is given.
[0004]
By the way, when comparing feature amounts as described above, there is a problem in that the reliability of comparison and collation decreases when the posture and facial expression of a person at the time of authentication are different from those at the time of registration. For this reason, for example, it is necessary to restrict the posture and facial expression, resulting in inconvenience that the person to be authenticated is uncomfortable. Therefore, conventionally, when calculating the feature amount at the time of registration, the feature amount of the posture and the expression is calculated by a predetermined evaluation formula and registered as a determination value, and at the time of authentication, the measured feature amount is calculated. By calculating using the same evaluation formula and comparing with the registered value for determination, the reliability of authentication was improved while reducing discomfort to the person to be authenticated (for example, patent document) 1).
[0005]
[Patent Document 1]
JP-A-6-119433 (FIG. 1)
[0006]
[Problems to be solved by the invention]
However, the visual characteristics of a person can change variously due to changes in lighting, changes in person's makeup, aging, and the like during registration and authentication. That is, there is a high possibility that inconvenience such as non-authentication of a registered person to be authenticated occurs, and there is a problem that it is not sufficient to consider only the posture and facial expression at the time of registration and authentication. Therefore, there is a desire to obtain a further index in order to obtain higher authentication reliability.
[0007]
The present invention has been made in view of the above problems, and an object of the present invention is to obtain a personal authentication system that can obtain higher authentication reliability by using a new index in authentication based on a face image.
[0008]
[Means for Solving the Problems]
In order to achieve the above object, the personal authentication system of the present invention includes registered surface image acquisition means for acquiring at least a face image of a person to be authenticated at the time of registration, and at least a facial skull of the person to be authenticated at the time of registration. Registered fluoroscopic image acquisition means for acquiring a fluoroscopic image, data storage means for storing authentication data of a registered person based on at least the registered surface image and the registered fluoroscopic image, and at least a face image of the face of the person to be authenticated at the time of authentication Registered based on the authentication surface image acquisition means for acquiring the similarity, the degree of similarity between the registered surface image and the authentication surface image, and whether or not there is an anatomical contradiction between the registered fluoroscopic image and the authentication surface image And an authentication unit for performing personal authentication between the authenticated person and the person to be authenticated.
[0009]
Here, the registered surface image acquisition means for acquiring a surface image is an image acquisition means using, for example, a CCD, and the registered fluoroscopic image acquisition means for acquiring a fluoroscopic image is, for example, an image using an X-ray or the like. It is an acquisition means. Further, the authentication surface image acquisition means for acquiring the surface image may be of the same configuration as the registered surface image acquisition means, for example, an image acquisition means using a CCD or the like. May be. Furthermore, the degree of similarity between the registered fluoroscopic image and the authentication surface image is, for example, a degree indicating whether or not the structure of the skull and the surface structure of the face are similar without any anatomical contradiction.
[0010]
According to this configuration, in addition to the comparison and collation between the registered surface image and the authentication surface image, a comparison and comparison between the registered fluoroscopic image and the authentication surface image is performed. Even when such a change occurs, a perspective image such as a facial cranium that is not easily affected by such a change is considered as one of the comparisons, so that the reliability of personal authentication can be improved.
[0011]
In order to achieve the above object, the personal authentication system of the present invention is characterized in that, in the above configuration, the registered surface image, the registered fluoroscopic image and the authentication surface image are three-dimensional stereoscopic images.
[0012]
According to this configuration, it is possible to easily increase the number of comparison / collation items and further improve the reliability of personal authentication.
[0013]
In order to achieve the object as described above, in the personal authentication system of the present invention, in the configuration described above, the authentication unit corresponds to at least one of the contours of the orbit, pear-shaped mouth, and skull of the registered fluoroscopic image and the authentication surface image. It is characterized by verifying whether or not there is an anatomical contradiction with the surface portion.
[0014]
Here, when an orbit is used as the registered fluoroscopic image, it is possible to compare and collate the position of the outer eye corner point or the inner eye corner point of the authentication image with respect to the eye socket, the position of the line connecting the outer eye corner point and the inner eye corner point, or the like. . Moreover, when using a pear-shaped mouth, the relationship with a nose wing can be compared and collated. Further, in the case of using the contour, for example, a curve of the frontal lobe (forehead portion) or the like can be obtained by acquiring a three-dimensional stereoscopic image for the registered fluoroscopic image. In this case, the subcutaneous thickness of the forehead is substantially constant, and can be easily compared with the surface curve obtained from the authentication surface image and the skull curve obtained from the registered fluoroscopic image.
[0015]
According to this configuration, it is possible to perform comparison and collation using the registered fluoroscopic image effectively, and the reliability of personal authentication can be further improved.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention (hereinafter referred to as embodiments) will be described with reference to the drawings.
[0017]
FIG. 1 shows a conceptual block diagram of a personal authentication system 10 of the present embodiment.
[0018]
Characteristic items of the present embodiment are registered surface image acquisition means for acquiring at least a face image of a person as a registered image acquired at the time of authentication person registration, and a registered perspective image for acquiring a perspective image of at least a facial skull of the person. Acquisition means, and feature information obtained from a fluoroscopic image is used as one of the verification means for personal authentication.
[0019]
A surface shape measurement sensor (for example, a CCD camera) 12 can be used as the registered surface image acquisition means, and a bone shape measurement sensor (for example, an X-ray camera) 14 can be used as the registered fluoroscopic image acquisition means. . The surface shape measurement sensor 12 acquires a front image (surface image) of a person to be registered, for example, a face, and supplies it to the surface feature information extraction unit 16. The surface feature information extraction unit 16 extracts, for example, the shape and size of a part that is a feature of a person such as eyes, nose, mouth, and eyebrows and the geometric positional relationship between the surface images according to a predetermined algorithm. The registration unit (data storage means) 18 registers. On the other hand, the bone shape measurement sensor 14 acquires a front image (perspective image) of the facial cranium of the person to be registered and supplies it to the bone feature information extraction unit 20. In the bone feature information extraction unit 20, for example, an orbit (an irregular quadrangular pyramid-shaped indentation that can hold the eyeball and its appendage) or a pear-shaped mouth (nasal bone and maxilla) according to a predetermined algorithm. The surface of the eyes, nose, mouth, eyebrows, etc. extracted from the surface image of the same person by extracting the outline of the skull as bone feature information, etc. The information is registered in the registration unit 18 in association with the feature information.
[0020]
In addition to this, general personal information of a registered person, for example, name, affiliation, ID number, and the like may be associated with the registration unit 18 and registered.
[0021]
The personal authentication system 10 is also provided with a surface shape measurement sensor (for example, a CCD camera) 22 as an authentication surface image acquisition unit used at the time of authentication. Similar to the surface shape measurement sensor 12, the surface shape measurement sensor 22 acquires a front image (surface image) of the person to be authenticated, for example, and supplies it to the surface feature information extraction unit 24. The surface feature information extraction unit 24 extracts, for example, the shape and size of a person's feature such as eyes, nose, mouth, and eyebrows and the geometric positional relationship between the surface images according to a predetermined algorithm. . Accordingly, the surface shape measurement sensor 22 and the surface feature information extraction unit 24 may be provided separately from those used during registration as shown in FIG. 1, or may be shared during registration and authentication.
[0022]
An authentication unit 26 is connected to the surface feature information extraction unit 24 and the registration unit 18. In the authentication unit 26, the feature information extracted based on the authentication surface image acquired by the surface feature information extraction unit 24 and the feature information registered in the registration unit 18 are sequentially compared and collated. It is confirmed whether or not the feature information exists as registered information, and the result is provided to the result processing unit 28. In the result processing unit 28, when the feature information of the person to be authenticated exists as registration information, for example, “authentication OK” is displayed and voiced, and the accompanying control, for example, gate passage permission and passage recording is performed. Perform processing. In addition, when the authentication target person's characteristic information does not exist as the registration information, “authentication NG” is displayed and voiced, and control associated therewith, for example, rejection of gate passage or rejection recording processing is performed.
[0023]
FIG. 2 is a flowchart showing the flow of processing when registering an authenticated person in the present embodiment. In the personal authentication system 10, when registering an authenticated person, the registration operation is preferably performed in the same posture as at the time of authentication, for example, when authentication is performed in a standing posture at the time of authentication. When authentication is performed with a sitting posture, the same chair or the like is used during registration and the registration operation is performed with the sitting posture.
[0024]
In the personal authentication system 10, when a registration start switch (not shown) is turned on, the surface shape measurement sensor (CCD camera) 12 operates to use an appropriate illumination or strobe, and as shown in FIG. A front image (surface image) 30a is taken (S100). Similarly, the bone shape measurement sensor (for example, X-ray camera) 14 operates to capture a front image (perspective image) 32 of the facial skull of the registered authentication person as shown in FIG. 4 (S101).
[0025]
Thereafter, each of the surface feature information extraction unit 16 and the bone feature information extraction unit 20 first normalizes the acquired image (S102). For example, the surface feature information extraction unit 16 performs rotation, enlargement, reduction, and the like of the acquired surface image 30a, and corrects the image size to be suitable for feature point extraction on the image area as shown in FIG. 3B. A registered surface image 30b is obtained. Similarly, the bone feature information extraction unit 20 also obtains a registered perspective image 32 that is corrected to an image size suitable for feature point extraction on the image region by rotating, enlarging, and reducing the obtained perspective image (see FIG. 4 shows a registered fluoroscopic image after normalization). At this time, as shown in FIG. 4, it is desirable to perform edge processing or the like on the registered fluoroscopic image 32 such as an X-ray image so that the shape of the facial skull becomes clear. This normalization process is performed in order to easily compare with the image acquired at the time of authentication. For example, in the case of a surface image, the position of the eyes is positioned at the reference position on the image area. To do. In the case of a fluoroscopic image, the top of the frontal bone is positioned at the reference position on the image area.
[0026]
Subsequently, the surface feature information extraction unit 16 and the bone feature information extraction unit 20 extract feature information for each registered authenticator from the acquired registered surface image 30b and registered fluoroscopic image 32, respectively (S103). As described above, the shape and size of eyes, nose, mouth, eyebrows and the like, and the geometric positional relationship between each other are extracted from the registered surface image 30b, and the eye socket or pear shape is extracted from the registered fluoroscopic image 32. The size, position, shape, etc. of the mouth and the like, the outline of the skull, etc. are extracted and registered in the registration unit 18 as registered face feature information and registered bone feature information in association with each other (S104).
[0027]
In the registration unit 18, the information of each registered authenticator is individually managed. For example, only information regarding eyes in which the eye information of the surface image and the information of the eye socket of the fluoroscopic image are associated with each other, You may create the information group which collected only the information regarding the nose which matched the pear mouth.
[0028]
FIG. 5 shows a flowchart for explaining an operation procedure at the time of authentication of the personal authentication system 10. In this embodiment, at the time of authentication, only the surface image (authentication surface image) of the person to be authenticated is acquired (S200). That is, by operating the surface shape measurement sensor (CCD camera) 22 and using an appropriate illumination or strobe, the front image (surface image) 30a of the person to be authenticated is photographed in the same manner as shown in FIG. . Then, the surface feature information extraction unit 24 performs normalization for performing rotation, enlargement, reduction, and the like of the acquired authentication surface image in the same manner as at the time of registration (S201), and extracts feature information, that is, eyes, nose, mouth, eyebrows, and the like. The shape, size, and mutual geometric positional relationship are extracted (S202), and the information is provided to the authentication unit 26.
[0029]
When the authentication unit 26 receives the feature information of the person to be authenticated from the surface feature information extraction unit 24, the authentication unit 26 compares it with all the registered face feature information registered in the registration unit 18, and calculates the similarity to each registered face feature information. (S203). For the calculation of the degree of similarity, any conventional comparison / collation method used can be used. In this case, for example, it is desirable to obtain similarities for eyes, nose, mouth, eyebrows, and the like, and further integrate the similarities to indicate the total similarity.
[0030]
Subsequently, the authentication unit 26 assigns authentication priorities in order from the highest similarity among the calculated similarities. That is, the registrants Sn = S1, S2,... SN (n = 1 to N) are ranked (S204), and actual authentication work is started.
[0031]
First, in starting the authentication work, the registrant S1 having the highest similarity is extracted (n = 1) (S205). The authentication unit 26 has a threshold (for example, 80%) for determining whether the similarity between the registered surface image and the authentication surface image exceeds a predetermined value, and whether the similarity exceeds the threshold. A determination is made (S206). At this stage, if the feature information of the authentication surface image does not have a similarity greater than or equal to the threshold with respect to the feature information of the registered surface image, the person who has requested authentication is registered in the personal authentication system 10. It is determined that there is no non-registered person (S207), and the authentication unit 26 instructs the result processing unit 28 to perform non-authentication processing. That is, non-authentication is indicated by a display or sound, and a series of personal authentication processes is terminated with a record of the non-authenticated person as necessary.
[0032]
As will be described later, in the present embodiment, the registrant Sn exceeding the above threshold (S1 in the above case) is set to an extremely high threshold for further authentication based on the bone feature information. There is no need to do this, and there will be a change in the surface due to changes in lighting, makeup of the person, aging, etc., and at this point it will contribute to the prevention of misjudgment that the registered person will be unauthenticated can do.
[0033]
On the other hand, if it is determined in step (S206) that the similarity of the feature information based on the surface image exceeds a predetermined threshold, the authentication unit 26 displays the fluoroscopic image corresponding to the registrant S1 having a similarity exceeding the threshold. (I.e., bone feature information) is called from the registration unit 18 to verify whether or not there is an anatomical contradiction with the authentication surface image acquired by the surface shape measurement sensor 22 (S208).
[0034]
Here, the point of interest in determining whether or not an anatomical contradiction exists is, for example, when the feature point of interest of the authentication surface image is “eye”, as shown in FIG. The position of the eye fissure c that connects the eye angle point (eye gashira) a and the external eye angle point (eye gill) b is normally set to 3/10 below the orbital height, but whether or not the condition is satisfied Or, when the feature point of interest of the authentication surface image is “nose”, as shown in FIG. 6B, the pear-shaped mouth B obtained from the registered fluoroscopic image is completely inside the nose wing d obtained from the authentication surface image. If the feature point of interest of the authentication surface image is “contour outline”, the skull curve of the forehead portion obtained from the registered fluoroscopic image does not intersect the surface curve of the forehead portion obtained from the authentication surface image Or not, whether or not the skull curve is outside the surface curve, and so on. In addition, when comparing a skull curve and a surface curve, the registration fluoroscopic image and authentication surface image to be acquired need to be a three-dimensional stereoscopic image or a side image.
[0035]
Therefore, when the above relationship is satisfied between the feature information based on the authentication surface image and the bone feature information based on the registered fluoroscopic image, that is, the authentication unit 26 sends the feature information of the person to be authenticated to the registration unit 18. If it is determined that there is no contradiction with the feature information based on the registered fluoroscopic image, the authentication unit 26 is the same person as the registrant S1 registered in the registration unit 18, that is, , The person having the characteristic information of S1 is authenticated (S209), and the result processing unit 28 is instructed to perform the authentication process. That is, it indicates that the authentication has been made by display or voice, and if necessary, a record of the authenticated person is left and the series of personal authentication processes is terminated.
[0036]
On the other hand, if it is determined in step (S208) that there is an anatomical contradiction, for example, the inner eye corner point a and the outer eye corner point b are located outside the eye socket A, or the position of the eye fissure c is the eye socket. In the case where there is a contradiction such as the nose wing d smaller than the pear mouth B, the skull curve and the surface curve intersecting, or the like, located at the upper part of A, the authentication unit 26 determines the characteristics of the registrant used this time. It is determined that the information is not that of the person to be authenticated, and in order to compare with other feature information, it is determined whether comparison authentication has been performed for all of the feature information registered in the registration unit 18 ( S210). If it is determined that there is an anatomical contradiction and it is determined that the verification for all registered information has been completed (n = N), there is no registration corresponding to the person to be authenticated, that is, non- It is determined that the user is a registrant (S207), and the authentication unit 26 instructs the result processing unit 28 to perform non-authentication processing. That is, non-authentication is indicated by a display or sound, and a series of personal authentication processes is terminated with a record of the non-authenticated person as necessary.
[0037]
On the other hand, if it is determined in step (S210) that all feature information collations have not been completed (n = N is not N), the next registrant S (n = n + 1) is called from the registration unit 18 (S211). ), The processing after step (S206) is performed, and it is sequentially determined whether the similarity exceeds a predetermined threshold value and whether the authentication surface image has an anatomical contradiction, as described above. The comparison of the feature points of the person who has requested authentication is performed.
[0038]
In this way, by performing two-stage comparison and collation using the registered surface image and the registered fluoroscopic image, for example, three registrants exhibiting a high degree of similarity exceeding the threshold in the comparison of only the surface image (S1, S2). , S3) If it exists, it is possible to screen three people by performing authentication using a fluoroscopic image. That is, when the similarity with the registrant S2 is the highest as a result of the authentication using the fluoroscopic image, the person to be authenticated can be determined to be S2.
[0039]
In this way, the authentication surface image acquired at the time of authentication is first authenticated for similarity using the registered surface image, and further, the authentication surface image is subjected to the condition at the time of image acquisition and the state of the subject's makeup. By performing similarity authentication using registered fluoroscopic images that are hardly affected by aging or the like, it is possible to suppress error factors during authentication and improve the reliability of personal authentication.
[0040]
FIG. 7 shows a flowchart showing another authentication procedure using a registered fluoroscopic image.
[0041]
In FIG. 7, the procedure of steps (S200) to (S202) for acquiring the authentication surface image of the person to be authenticated and extracting the feature information is the same as the procedure shown in FIG. In the case of the example in FIG. 7, when acquisition of the authentication surface image is completed, first, the authentication surface image is compared with all registered transmission images (bone feature information), and there is an anatomical contradiction between the two. Only registered transparent images that are not to be selected are selected (S300).
[0042]
Subsequently, the authentication surface image of the person to be authenticated (feature image of the person to be authenticated) and the registered surface image (registered face feature information) corresponding to the bone feature information selected in step (S300) are compared and collated. The similarity is calculated for each image (S301). The registrant corresponding to the highest similarity among the calculated similarities is set as a registrant Smax (S302).
[0043]
Subsequently, the authentication unit 26 determines whether or not the similarity of the registrant Smax exceeds a predetermined threshold (for example, 80%) (Smax> threshold) (S303). If the threshold value is exceeded, the authenticating unit 26 determines that the person to be authenticated has characteristic information of the registrant Smax and that the person to be authenticated is a registrant (S304). Then, the result processing unit 28 is instructed to perform authentication processing. That is, it indicates that the authentication has been made by display or voice, and if necessary, a record of the authenticated person is left and the series of personal authentication processes is terminated. On the other hand, if Smax> threshold is not satisfied in step (S303), it is determined that there is no registration corresponding to the person to be authenticated, that is, it is a non-registered person (S305). An instruction is issued to perform non-authentication processing. That is, non-authentication is indicated by a display or sound, and a series of personal authentication processes is terminated with a record of the non-authenticated person as necessary.
[0044]
In this way, for the authentication surface image acquired at the time of authentication, first, a registered fluoroscopic image that is hardly affected by the conditions at the time of image acquisition, the makeup state of the person to be authenticated, aging, etc. is used for the authentication surface image. Authentication is performed, and registration candidates with no anatomical contradiction are selected, and then the similarity is verified using the registered surface image, thereby efficiently suppressing the error factor at the time of authentication. Personal authentication can be performed to improve the reliability.
[0045]
Note that the conceptual configuration block diagram shown in FIG. 1 of this embodiment and the processing procedures shown in FIGS. 2, 5, 7, etc. are examples, and when performing personal authentication, at least the face of the registrant at the time of registration is shown. A configuration in which a surface image and a fluoroscopic image of the skull are acquired, and feature information obtained from the images is compared with feature information obtained from at least the face image of the face of the person to be authenticated acquired at the time of authentication. If there is, it can be changed as appropriate, and the same effect as in the present embodiment can be obtained.
[0046]
In the present embodiment, the feature points to be extracted from the fluoroscopic image have been described by taking the orbit, pear mouth, the outline of the skull, etc. as an example, but the position in the fluoroscopic image and the position in the surface image have relevance as characteristic points. If it is a thing, it can utilize suitably and the same effect can be acquired.
[0047]
【The invention's effect】
According to the present invention, in addition to the comparison and collation between the registered surface image and the authentication surface image, by performing the comparison and collation between the registered fluoroscopic image and the authentication surface image, a surface change in the person due to lighting, makeup, aging, etc. Even in the case of occurrence of such a problem, a fluoroscopic image such as a skull that is not affected by such a change is considered as one of the comparisons, so that the reliability of personal authentication can be improved.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration concept of a personal authentication system according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining an operation procedure at the time of registration of the personal authentication system according to the embodiment of the present invention.
FIG. 3 is an explanatory diagram for explaining acquisition and normalization of a registered surface image of the personal authentication system according to the embodiment of the present invention.
FIG. 4 is an explanatory diagram illustrating acquisition of a registered fluoroscopic image of the personal authentication system according to the embodiment of the present invention.
FIG. 5 is a flowchart illustrating an operation procedure during authentication of the personal authentication system according to the embodiment of the present invention.
FIG. 6 is an explanatory diagram illustrating an authentication method using a registered transparent image of the personal authentication system according to the embodiment of the present invention.
FIG. 7 is a flowchart illustrating another operation procedure during authentication of the personal authentication system according to the embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Personal authentication system, 12, 22 Surface shape measurement sensor, 14 Bone shape measurement sensor, 16, 24 Surface feature information extraction part, 18 Registration part, 20 Bone feature information extraction part, 26 Authentication part, 28 Result processing part.

Claims (3)

Registered surface image acquisition means for acquiring at least a face image of the face of the person to be authenticated at the time of registration;
Registered fluoroscopic image acquisition means for acquiring a fluoroscopic image of at least the facial skull of the person to be authenticated at the time of registration;
Data storage means for storing authentication data of a registered person based on at least the registered surface image and the registered fluoroscopic image;
Authentication surface image acquisition means for acquiring at least the surface image of the face of the person to be authenticated at the time of authentication;
Personal authentication between a registered person and a person to be authenticated based on the similarity between the registered surface image and the authentication surface image, and verification of whether or not there is an anatomical contradiction between the registered fluoroscopic image and the authentication surface image An authentication unit that performs
A personal authentication system comprising: The system of claim 1, wherein
The personal authentication system, wherein the registered surface image, the registered fluoroscopic image, and the authentication surface image are three-dimensional stereoscopic images. The system according to claim 1 or claim 2,
The authentication unit is configured to verify whether or not there is an anatomical contradiction between at least one of the contours of the orbital, piriform mouth, and skull of the registered fluoroscopic image and the corresponding surface portion of the authentication surface image. Personal authentication system.

Images