JP2005149527A - System and method of facial authentication - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve convenience by enabling a adjustment balance between performance of personal authentication and conformity, and strength to a variation of environment depending on the condition. <P>SOLUTION: A system for facial authentication comprising a facial image input 41 for inputting a facial image of a user, a feature extractor 42 for extracting feature information from the facial image 42, 43, an authentication information storage 45 for storing corresponding pairs of registered facial image information for authentication and identification information inherent to the user, a collator 44 for collating the feature information with the registered information for authentication, an identification information acquisition part for acquiring the identification information from the user, and a re-collation controller for making the collator re-collate the registered facial image information for authentication, extracted from the authentication information storage 45 corresponding to the identification information acquired by the identification information acquisition part, with the feature information, when the collator, as a result of collation by the collator, cannot specify the registered information for authentication corresponding to the feature information. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a face authentication device and a face authentication method using a face image.

  So far, authentication technologies for authenticating individuals from biometric information have been studied in various fields with the aim of realizing higher security. In particular, a personal authentication system based on face image recognition using image processing technology has been attracting attention because of its comfort in that little time or effort is required for a target person to input biometric information necessary for authentication. However, although facial image recognition is superior in comfort to recognition using biometric information such as other irises and fingerprints, it is inferior in recognition performance, which has been an obstacle to application to a personal authentication system. . In order to achieve high security using facial image recognition, both the rate of identifying the person (passing rate of the person) and the rate of rejecting persons other than the person (exclusion rate of others) are not sufficiently high. must not.

  However, in the current face image recognition, the image of the person to be authenticated and the registered information are compared and collated in a predetermined feature space to obtain the similarity, and the person to be authenticated is determined whether or not the degree of similarity exceeds a predetermined threshold. It is common to determine whether or not a person is registered information, but if a threshold value that is a criterion for determination is fixed, there is an inconvenience that a sufficient height cannot be obtained at least in one of the aforementioned rates. Occur.

  For example, the relationship between the exclusion rate of others and the passing rate of the person is as shown in FIG. 20, and if the threshold is set high as shown in (A), the rate of passing the other person is lowered, but the person may be rejected at the same time. If the threshold is set low as in (B), the person can always pass, but others can also pass. That is, it has been difficult for a conventional personal authentication system using face image recognition to sufficiently achieve both high security and comfort including that a regular user is not rejected. Note that the comfort in inputting biometric information does not reach the face image, but the same problem as described above also occurs in the authentication for voice as biometric information.

  As described above, it is difficult to keep both the pass rate and the exclusion rate of people with conventional biometrics authentication using faces and voices. In addition, including the strength against environmental conditions such as lighting and noise, It was difficult to realize a personal authentication system with sufficient authentication performance and comfort using only one authentication logic.

  The present invention has been made in view of such problems, and its purpose is to make it possible to adjust the balance between the performance of personal authentication and comfort, the strength against environmental fluctuations, etc. according to the situation, It is to provide a system with improved convenience.

  In order to achieve the above object, in the present invention, a face image input unit that inputs a face image of a person to be authenticated, a feature extraction unit that extracts feature information from the face image, face image registration information for authentication, and the person to be authenticated An authentication information storage unit that stores the identification information unique to each other, a collation unit that collates the feature information with the authentication-required registration information, an identification information acquisition unit that acquires the identification information from the person to be authenticated, and As a result of the collation performed by the collation unit, when the authentication registration information corresponding to the feature information cannot be specified, authentication face image registration information corresponding to the identification information acquired by the identification information acquisition unit is obtained. A re-collation control unit that is extracted from the authentication information storage unit and causes the collation unit to perform re-collation using the extracted authentication face image registration information and the feature information. is there.

  Further, in order to achieve the above object, in the present invention, a storing step of storing authentication target face image registration information of a person to be authenticated in association with ID data unique to the person to be authenticated, and a face image of the person to be authenticated A first input step for input, a first verification step for verifying the face image of the person to be authenticated input in the first input step and the face image registration information for authentication, and in the first verification step If not authenticated, a second input step for inputting the ID data and face image of the person to be authenticated, and the authentication corresponding to the ID data input in the second input step from the authentication required face image registration information And a second collation step for extracting the facial image registration information and re-collating the registration information with the face image of the person to be authenticated input in the second input step. is there.

  According to the present invention, it is possible to control the balance between safety and comfort by using a conventional biometric authentication method that has been difficult to realize a security system that combines safety and comfort at a high level. A new system that realizes convenience can be provided. In the present invention, it is possible to provide a system that comprehensively prevents a decrease in recognition rate by detecting an environmental change around the person to be authenticated and selecting an optimal authentication method according to the change.

  First, embodiments of the biometric information authentication apparatus and method according to the present invention will be described with reference to the drawings. Here, an embodiment of authentication using a face image as biometric information will be described.

  FIG. 1 shows a block format showing the basic structure of the biometric information authentication apparatus according to the present invention. The biometric information receiving unit A is a means for acquiring input biometric information necessary for authentication from the person to be authenticated, and has three basic components: an image input unit 1, a face detection unit 2, and a feature extraction unit 3. The image input unit 1 is a processing unit that acquires an image of a certain area including the face image of the person to be authenticated and transmits the input image information to the face detection unit 2 as digital data. The face detection unit 2 analyzes the input image information sent from the image input unit 1 to determine whether or not a face image is included. If it is included, the face detection unit 2 extracts the face image information and extracts a feature extraction unit. 3 is a processing means to send to 3. The feature extraction unit 3 is a processing unit that calculates the feature amount in the face image in the input image information by the feature extraction algorithm selected by the mode management unit C, and sends the result to the matching unit 4 as biometric feature information.

  The identification unit B is means for determining whether the person to be authenticated is included in the registered person and in which category it is classified. The identification unit B includes a verification unit 4 and an authentication information storage unit 5. The collation unit 4 collates the biometric feature information sent from the biometric information receiving unit A with the registered registration information for authentication using the identification method determined by the mode control information sent from the mode management unit C, and determines the similarity. The processing means for determining whether the person to be authenticated matches one of the registered persons. The authentication information storage unit 5 is processing means for storing registration information for authentication used when the collation unit 4 obtains the similarity. When the collation unit 4 performs identification, the authentication information storage unit 5 stores the authentication registration information. To transmit.

  The mode management unit C is a means for managing / controlling the identification method used when the verification unit 4 compares and identifies the biometric feature information of the person to be authenticated with the registered information, and determines the identification method based on various information. Then, it has processing means for transmitting the result to the collation unit 4 as mode control information, that is, a mode determination unit 6. As a result, the system controls authentication performance and comfort, and can improve the convenience of the person to be authenticated.

  Examples of authentication logic management / control patterns by the mode management means include the following five.

(1) Mode control by time.

(2) Mode control based on information sensed by the sensor.

(3) Mode control based on environmental information included in the input image.

(4) Mode control by external operation by the administrator and the person to be authenticated.

(5) Control of a mode for selecting which of a plurality of types of different biological information is to be acquired from the person to be authenticated for use in authentication.

(6) Control of a mode for selecting which of biometric information and non-biometric authentication information to acquire from the person to be authenticated for use in authentication.

  (1) to (3) are systems in which the biometric information authentication apparatus has some means for automatically collecting information for controlling the mode, and the mode management unit C controls the mode based on the information. is there. Such a mode control method is called automatic mode control. (4) is characterized in that the biometric information authentication apparatus includes an external input device for sending an instruction for the manager and the person to be authenticated to instruct the mode management unit C to control the mode. . The mode management unit C determines the mode based on the information sent from here and transmits it to each unit. Such a mode control method is called manual mode control. (1) to (4) are examples in which the mode management unit pays attention to information that is an element for determining the mode, but (5) and (6) In this embodiment, whether the input information is acquired is controlled by the mode management unit. Examples of the authentication mode control mechanisms (1) to (6) will be described.

(Example 1)
The first embodiment is the time mode control (1). In this case, when the preset time comes, the mode management means issues a command (mode control information) for switching to a certain identification method different from that to the identification means. That is, the identification means can perform identification work using a different identification method depending on a preset time zone.

  Here, a biometric information authentication apparatus incorporating the time mode control system will be described.

  Fluctuations in the pass rate of others and the exclusion rate of others with respect to the threshold for determining authentication (FIG. 20: Osamu Yamaguchi, Kazuhiro Fukui: “Face recognition system using moving images”, PRMU 97-50, pp. 17-24 (1997) As can be seen from the above, it is impossible to fix both of the thresholds high enough wherever the threshold is set. Therefore, two thresholds that can obtain a sufficient height for either one are set. In this embodiment, the threshold setting 1 (person's passing rate is high but the other person's exclusion rate is slightly low) and the threshold setting 2 (safety priority mode) is set. Prepare two.

  By changing these modes according to the time of day, a security system that emphasizes comfort during the day and safety during the night is realized. (FIG. 2) In the time control mode, the mode manager C can control the threshold necessary for identifying the person in two ways according to the time zone. The threshold setting 1 is used in the daytime when high safety is not required, and the threshold setting 2 is used in the night when it is necessary to obtain sufficient safety. This is achieved by the mode manager C changing the threshold value without intentional operation by a human, and such mode control is called automatic mode control. The processing flow of the security system realized by using these two threshold settings properly will be described with reference to the flowchart of FIG.

  The image input unit 1 acquires image information using a camera installed in consideration of capturing the face of the person to be authenticated, and transmits the image information to the face detection unit 2. The face detection unit 2 sends the detected face image information to the feature extraction unit 3 only when a face is detected in the input image (step S1). The feature extraction unit 3 analyzes the sent face image information and sends the feature quantity acquired as biometric feature information to the matching unit 4. The mode determination unit 6 issues a command to the verification unit 4 to perform identification in the personal convenience mode if the operating time zone is noon, and in the safety priority mode if it is night (step S2). The collation unit 4 obtains the feature similarity by collating both the biometric feature information sent from the feature extraction unit 3 and the authentication registration information sent from the authentication information storage unit 5, and the largest of the similarities is the mode. Identification is performed based on whether or not the threshold setting selected in accordance with the instruction from the determination unit 6 is exceeded, and the result is output (step S3). In this embodiment, switching between two modes has been described. However, the number of modes is not limited to two, and a plurality of three or more modes may be used. The same applies to the following Examples (2) to (4).

(Example 2)
Face image recognition is not sufficiently robust against variations in lighting conditions, and a feature extraction algorithm that is robust against variations in lighting has a problem that authentication performance under standard lighting conditions is lower than those that do not. Therefore, an illuminance sensor is newly provided, the intensity of illumination is sensed by this sensor, and a mode control example corresponding to the above (2) for switching the feature extraction algorithm to one that is robust to illumination according to the intensity is introduced. An example of realizing such mode control is a second embodiment according to the present invention which will be described next.

  The apparatus according to the present embodiment configures a biometric information authentication apparatus that is robust to illumination by using a face image as an authentication item and using two feature extraction methods having different properties with respect to illumination conditions. The two feature extraction methods are “algorithm 1 that can achieve a sufficiently high recognition rate at the reference illumination intensity, but otherwise poor recognition rate”, and “recognition from algorithm 1 at the reference illumination intensity. “Algorithm 2 that can reduce the rate, but otherwise allow higher recognition rates”. The authentication when each of them is used is set to the normal mode and the illumination robust mode. (FIG. 4) Here, algorithm 1 and algorithm 2 will be described.

  A face image obtained by photographing a person's face under a certain illumination condition is a reflection luminance image created by reflecting illumination light to a three-dimensional object called a face. The apparatus inputs such a reflected luminance image of the face, normalizes it to a predetermined size, and further normalizes the luminance range to form an image pattern. The algorithm 1 is a feature extraction algorithm that forms an image pattern based on the reflected luminance expression generated in this manner as biometric feature information.

  However, the image pattern based on the reflected luminance expression is object deformation (changes in three-dimensional shape such as facial expression changes), illumination fluctuations (changes in illumination spectral intensity and illumination direction due to time zone and weather), and object posture fluctuations (face orientation). Changes in the imaging direction such as changes in image quality). In particular, regarding the illumination direction, it is difficult to pre-learn patterns for various illumination directions in the verification registration information in spite of the unstable difference that the shading of the image pattern changes dramatically. Unlearned illumination fluctuations give unlearned illumination brightness to all input patterns. For this reason, algorithm 1 causes a serious reduction in recognition rate under lighting conditions different from those during learning.

  In order to prevent such a decline in the recognition rate due to illumination fluctuations, recognition is performed using biometric feature information based on feature quantities (ie, image representations) that are not easily affected by illumination and that preserve feature information unique to an object (personal face). Need to do. Since the illumination variation gives a strong variation to the reflected luminance image of the object, the influence of the illumination variation is inevitable as long as the reflected luminance expression is the biometric feature information. On the other hand, it is known that a certain kind of image representation derived from the reflected luminance pattern has a constancy that is hardly affected by illumination conditions.

  For example, in the literature (Yael Moses et al., “Face Recognition: the Problem of Compensation for Changes in Illumination Direction”, Proceeding of ECCV '94, 1994), an effective feature for the luminance variation is an image with respect to luminance variation. A brightness gradient expression (hereinafter referred to as a differential image) obtained in this way has been proposed.

  The first-order differentiation to the reflected luminance image corresponds to an operation for obtaining the luminance change rate, and a portion (edge) where the luminance changes stepwise is extracted. The object surface has a pattern of high and low reflectivity specific to the object, and the low reflection brightness always changes regardless of the illumination intensity at locations with low reflectivity, and the reflection brightness changes depending on the illumination intensity at locations with high reflectivity. To do. For this reason, the first-order differential image represents (a) a change in reflectance at a location where illumination is applied and (b) a change in shadow at a location where high reflectance is applied. (A) is an edge where there is no fluctuation in position even if there is an intensity fluctuation due to illumination fluctuation unless the attitude of the object changes, and (b) is an intensity and position which both fluctuate due to illumination fluctuation even if the attitude of the object does not change. It is an edge.

  The slope of the edge extracted by differentiation depends on the nature of the operator in the differentiation process. Generally, a smoother edge can be extracted as the operator is larger, and only a sharper edge is extracted as the operator is smaller. Since what differential operator is used depends on the contrast intensity due to illumination, etc., in this embodiment, the mode is combined with control for reducing the size of the differential operator used in accordance with the magnitude of the illumination intensity. Performed by the decision unit.

  The first effect of the differential image with respect to illumination fluctuation is that (a) preserves the characteristic of the reflectance pattern unique to the object according to the posture of the object. This shows the validity of using the differential pattern for object recognition. The second effect of the differential image on the illumination variation is that (a) appears stronger than (b) as the height difference of the reflectance pattern on the object surface increases, that is, the texture of the object is stronger. In particular, when the object has a smooth curved surface, the shadow due to illumination shows a gentle change, so that (b) is further weakened. This means that an unlearned illumination variation that was a large-scale variation in the reflected luminance expression is suppressed to an unlearned variation relating to (b) in the differential image (localization and miniaturization). The differential image is considered to have illumination variation constancy due to such an effect. The algorithm 2 receives a face reflection luminance image, differentiates it, normalizes it to a predetermined size, and further normalizes the luminance range to obtain an image pattern. As a result, in the algorithm 2, the influence of the illumination variation is localized and miniaturized, and the illumination variation robustness is improved.

  However, algorithm 2 drops the shadow information on the forehead and cheeks (reflecting the three-dimensional features of the individual's face) and uses only the texture information such as eyebrows and eyes as personality. This is because the recognition rate is slightly inferior to Algorithm 1 which can recognize personality as a clue. As a result, algorithm 1 is an algorithm that “can achieve a very high recognition rate under lighting conditions when learning registration information for verification, but when the lighting conditions deviate from this, the recognition rate extremely deteriorates”, and algorithm 2 The algorithm is “a reasonably high recognition rate can be achieved under illumination conditions when learning registration information for verification, and the recognition rate is stable even if the illumination conditions change”. Note that variations in illumination intensity are absorbed by normalizing the brightness of the image pattern in any algorithm.

  FIG. 5 shows a block configuration of the present embodiment. The structural diagram is very similar to FIG. 1, but there are two different parts. First, a device for sensing lighting conditions around the person to be authenticated is attached. The illumination sensor unit 17 is processing means for transmitting the illumination intensity in the target range to the mode determination unit 16 as digital data (illumination intensity information). Secondly, the mode determination unit 16 is controlled by the mode extraction unit 13 instead of the collation unit 14. In order to create a system that is resistant to illumination, the algorithm used for feature extraction must be changed to one that is robust to illumination, and must act on the feature extraction unit 13.

  Based on the above, the processing flow of the apparatus of this embodiment will be described with reference to the flowchart of FIG. First, as in the first embodiment, face image information is sent to the feature extraction unit 13 by the face detection unit 12 that extracts the facial features from the input image acquired by the image input unit 11 (step S11). The illumination sensor unit 17 always senses the illumination intensity around the person to be authenticated, and transmits illumination intensity information to the mode determination unit 16. When the mode determining unit 16 determines that the current illumination intensity does not change much from the illumination intensity at the time of learning the registered information for verification and is within a range that is sufficiently acceptable by the algorithm 1, the change in the illumination intensity is too large. If it is determined that a command is issued to the feature extraction unit 13 to adopt the illumination robust mode each time (step S12). The feature extraction unit 13 selects an algorithm according to this and performs feature extraction work (step S13). This allows the system to achieve an overall higher recognition rate than using each mode alone. The biometric feature information generated by the feature extraction unit is sent to the collation unit 14, where authentication is performed by collating with the registration information for authentication sent from the authentication information storage unit 15.

As a modification, it is possible to change the identification algorithm used in the collation unit 14 for each feature amount generated in the feature extraction unit 13 by changing the mode. (Fig. 7)
(Example 3)
Similarly to the second embodiment described above, a biometric information authentication apparatus that is robust to illumination conditions can be implemented using the mode control example (3) corresponding to the above (3). The basic block configuration is as shown in FIG. In this apparatus, an illumination intensity analysis unit 27 is newly installed between the face detection unit 22 and the feature extraction unit 23 instead of the illumination sensor as compared with FIG. 5 of the second embodiment. The illumination intensity analysis unit 27 analyzes the face image information input from the face detection unit 22, and sends, for example, the average luminance of the face image information to the mode determination unit 26 as illumination intensity information. When receiving the illumination intensity information, the mode determination unit 26 determines a feature extraction algorithm and instructs the feature extraction unit 23 to determine the feature extraction algorithm. The face image information acquired by the face detection unit 22 is identified by the collation unit 24 via the feature extraction unit 23. At this time, the feature extraction algorithm used by the feature extraction unit 23 is the same two modes, the normal mode and the illumination robust mode, as in the second embodiment.

  Next, the flow of processing in this apparatus will be described. Although the subject of each processing step is different, the flowchart is the same as FIG. 6 used in the second embodiment. First, the image input unit 21 acquires an image of a predetermined area and sends it to the face detection unit 22 as input image information. When the face detection unit 22 detects a face in the input image, the face detection unit 22 extracts and outputs face image information (step S11). The face image information is transferred to the illumination intensity analysis unit 27 as it is, and, for example, the average luminance in the image is calculated as the illumination intensity. The illumination intensity information is sent to the mode management unit C and becomes an element for determining an optimum mode. The mode determination unit 26 in the mode management unit C determines an algorithm to be used for feature extraction based on the received illumination intensity information, and transmits the result to the feature extraction unit 23 as mode control information (step S12).

  An instruction is given to select the normal mode if the illumination intensity does not change much from the illumination intensity at the time of learning the registration information for verification and is sufficiently within the allowable range by algorithm 1, and the illumination robust mode if the change in illumination intensity is too great. The feature extraction unit 23 extracts features from the face image information by the face detection unit 22 using an algorithm determined according to the mode control information (step S13), and sends the acquired biometric feature information to the matching unit 24. In the identification unit B, the verification unit 24 performs authentication using the authentication registration information sent from the authentication information storage unit 25 and the biometric feature information sent from the feature extraction unit 23. By such a system, the person to be authenticated can receive the personal authentication robust to the lighting as in the case of the second embodiment.

  The three examples given so far have all been automatic mode control in which the mode manager B automatically determines the mode based on information obtained by any functional block in the apparatus. Unlike the above, the fourth embodiment described below can control the mode adopted by the system by sending a signal to the mode manager B to operate the mode by the manager or the person to be authenticated. . Such mode control is referred to as manual mode control.

Example 4
Here, the mode control corresponding to the above (4) is adopted, and the manager or the person to be authenticated inputs a signal (mode selection information) for selecting the mode to the mode manager C from the outside, and according to it. A security system (apparatus a) and (apparatus b) using a biometric authentication apparatus that determines a mode to be adopted will be described. Regarding the mode control by the person to be authenticated in (device b), an embodiment will be described in which non-biometric authentication information using physical means such as storage of a password, an ID card or the like is used in addition to biometric information. In the following, a system (apparatus a) and (apparatus b) provided with two identification methods (threshold settings 1, 2) used in the first embodiment will be described as embodiments of the apparatus.

  First, as an example of the device a controlled by the administrator, the user's convenience mode that emphasizes comfort is used while the administrator is in the vicinity of the device, for example, and the safety priority mode that emphasizes safety is adopted when leaving the reception. The system to perform will be described. This is based on the premise that sufficient fraud deterrence can be obtained even in the personal convenience mode because the manager checks the person to be passed when the manager is at the reception.

  The block structure is shown in FIG. The difference from the first embodiment is that the mode determination unit 36 does not control the mode according to the time zone, but instead determines the mode according to the signal from the manager external input unit 39. The manager external input unit 39 is a processing unit that transmits a signal input by the manager to the mode determination unit 36. The administrator can obtain the convenience of controlling the mode according to his / her convenience by sending a signal (mode selection information) for switching the mode.

  The flow of processing when this apparatus is used will be described with reference to the flowchart of FIG. Normally, this apparatus is controlled by the mode determination unit 36 so that the personal convenience mode with a low threshold is used when the verification unit 34 performs authentication. When the manager leaves the reception, the manager inputs a signal for switching from the personal convenience mode to the safety priority mode to the mode determination unit 36 using the manager external input unit 39. Upon receiving this signal, the mode determination unit 36 issues an instruction to switch the authentication mode to the safety priority mode with a high threshold (mode control information). The subsequent information transmission process is the same as in the first embodiment. The input image information acquired by the image input unit 31 is sent to the face detection unit 32. Therefore, when it is recognized that the face is included in the image (step S31), the feature extraction unit 33 sends the biometric feature information obtained by analyzing the face image information to the matching unit 34. Receiving this, the collation unit 34 performs authentication in the safety priority mode as instructed previously (step S32) (step S33).

  If the manager who has returned to the reception sends a signal to the mode management unit C so as to switch the authentication mode to the user convenience mode again, the original state can be restored.

  Next, the apparatus b will be described. As an example of this apparatus, a person to be authenticated can specify his / her registration authentication information by a personal ID such as a wireless tag, an ID card, and a password, and can perform one-to-one personal authentication using the input face image information. A security system using a biometric information authentication device will be described. A basic configuration of the apparatus b is shown in FIG. This system normally operates in a safety priority mode, and some authenticated users may be rejected even though they are registered. In such a case, the person to be authenticated inputs a personal identification number to the mode determination unit 46 using the external input unit 49 for the person to be authenticated, and selects a specific individual designated by the personal identification number with a high pass rate. Authentication can be received again after switching to the personal identification mode to be verified. As a result, it is possible to realize a very convenient security system in which the probability that the authenticated person who is the registered person is finally rejected is kept low while maintaining sufficiently high security.

  The difference from the configuration of the device a is that the external input device is a device for inputting a personal ID for the person to be authenticated, and an external input unit 49 for the person to be authenticated. The flow of processing when using this system will be described. Since the case where the normal safety priority mode is operating is described in the first embodiment, it will be omitted, and the case where the registered person is rejected (step S41) will be described with reference to FIG. To do. At this time, the person to be authenticated inputs the personal ID to the mode determination unit 46 using the external input unit 49 for the person to be authenticated, and the mode determination unit 46 sets the authentication mode to the personal identification mode only for the collation unit 44 this time. An instruction to change and the input personal ID are output (mode control information). The image input unit 41, the face detection unit 42, and the feature extraction unit 43 input the biometric information of the person to be authenticated again, and the acquired biometric feature information is sent to the collation unit 44. The collation unit 44 collates the biometric feature information with the single personal authentication information sent from the authentication information storage unit 45 in the personal identification mode, and outputs the result again (step S42).

  The authentication work described so far in the embodiments has been an identification problem for determining whether or not the person to be authenticated matches someone among many registered persons. On the other hand, the process after step S41 (backup mode) results in an identification problem for determining whether or not the person to be authenticated matches one of the registered persons. By having a backup process due to an identification problem with few misrecognitions, the security system in the present embodiment can realize very high safety.

(Example 5)
Unlike the biometric information used for authentication in the mode control examples (1) to (4), which is only one type (face image), the biometric information authentication device described in this embodiment corresponds to the above (5). Mode control is used, and two types of face image and iris are adopted. Further, in (1) to (4), all are mechanisms for changing the mode by a single means, but in this embodiment, a mechanism in which the modes are controlled by a plurality of means is adopted.

  The biometric information authentication device in this embodiment employs facial image authentication that emphasizes comfort during the day, and an iris that emphasizes safety in consideration of the necessity of vigilance since there is no administrator at night. The authentication used is adopted. Iris authentication requires having to look into the camera to acquire the iris image, and it is difficult to say that comfort is high, but the recognition rate is considered to have reached a level that can be applied to security so far. Yes. The mode used in the present embodiment will be described with reference to FIG.

  As described above, the mode used in the apparatus of this embodiment changes depending on the time zone. In the daytime, the mode control system by the administrator used in the apparatus a of Example 4 is adopted. This is a system that uses the user convenience mode that emphasizes comfort when the administrator is at the reception, and switches to the safety priority mode that emphasizes safety using an external input device when the administrator leaves the reception. Since it has already been described in the fourth embodiment, details are omitted. At night, the iris authentication mode is used in which authentication is performed using only the iris. In order to perform authentication using an iris image, this biometric information authentication apparatus has an imaging means for acquiring an iris image separately from that for acquiring a face image, and a function for detecting an iris from the image and extracting a feature amount. I have.

  The structure of the biometric information authentication apparatus according to the present embodiment will be described with reference to the block diagram FIG. The face image information receiving unit A and the face recognizing unit B having functions related to authentication using face images are omitted because the functions are the same as those of the biometric information receiving unit A and the recognizing unit B of the third embodiment. The iris image information receiving unit D includes an iris image acquisition unit 58, an iris detection unit 59, and an iris feature extraction unit 60. The iris image acquisition unit 58 is a processing unit that acquires an image by an imaging unit set so as to acquire an iris image of the person to be authenticated, and sends it to the iris detection unit 59 as input image information. The iris detection unit 59 determines whether or not an iris is included in the image by analyzing the transmitted input image information, and if detected, sends an image of the iris part to the iris feature extraction unit 60 Means. The iris feature extraction unit 60 is a processing unit that analyzes the transmitted iris image, extracts a feature amount for calculating the similarity in the iris collation unit 61, and sends it as iris feature information. The iris identification unit E is composed of an iris collation unit 61 and an iris information storage unit 62. The iris matching unit 61 calculates the similarity by comparing the iris feature information sent from the iris feature extracting unit 60 and the registered iris information sent from the iris information storage unit 62, and compares the iris feature information with the set threshold value for identification. It is a processing means for performing and outputting the result. The iris information storage unit 62 is processing means for storing registered iris information used when collation is performed in the iris collation unit 61 and transmitting it when necessary. The mode management unit C includes a mode determination unit 56 and an administrator external input unit 57. The mode determination unit 56 issues a command to use face authentication at daytime and iris authentication at night according to the time zone, and when the time belongs to daytime, the mode selection sent from the manager external input unit 57 This processing means determines whether to use the user convenience mode or the safety priority mode according to the information, and transmits the result to the face collation unit 54. If the time zone is noon, the iris information receiving unit D and the iris discriminating unit E are instructed to stop the function, and the mode is determined according to the mode selection information from the manager external input unit 57, and the face matching unit 34. If the time zone is night, the face image information receiving unit A and the face identifying unit B are instructed to stop the function, and the system is shifted to a system that performs only iris authentication. When the time zone is daytime, the manager external input unit 57 normally operates the personal convenience mode, but when the manager leaves the reception, the mode determination unit 56 sends a signal (mode selection information) for switching to the safety priority mode. And has the same function as the administrator external input unit 39 described in the fourth embodiment.

  The flow of processing when this biometric information authentication apparatus is used will be described with reference to FIG. First, assume that it is used during the daytime (step S51). At this time, the function related to iris authentication, the iris image information receiving unit D, and the iris identifying unit E are stopped by the mode determining unit 56, and only authentication using a face image is possible. When a face is detected by the face detection unit 52, face image information is output to the feature extraction unit 53 (step S52), and the face feature extraction unit 53 extracts a feature amount from the face image received from the face detection unit 52 (step S52). S53). The mode determination unit 56 uses the personal convenience mode when the manager is in the reception, and when leaving the reception, the administrator switches the mode using the external input unit 57 and performs the identification work in the safety priority mode ( In step S54), an instruction is issued to the face matching unit 54. Details of the face image authentication by the mode control of the administrator have been described in the fourth embodiment, and are omitted here. Next, the case where it uses at night is demonstrated (step S51). The function related to face authentication, the face image information receiving unit A, and the face identifying unit B are stopped by the mode determining unit 56, and only authentication using an iris image is possible. When the iris detection unit 59 detects an iris in the input image, the iris feature extraction unit 60 extracts a feature amount from the input iris image (step S56), and the iris collation unit 60 uses the feature amount and the registered iris information. Identification is performed (step S57).

(Example 6)
The present invention was originally intended to realize security with a high balance between safety and comfort, and was intended to overcome the problems of conventional recognition technology. Therefore, in this embodiment, the mode control example corresponding to the above (6) is used, and biometric information authentication is combined with non-biometric authentication information such as a wireless tag and a personal identification number so that the convenience of the person to be authenticated is reduced as much as possible. Explain the security system aiming at the sex.

  In the security system in the present embodiment, a face image is used as biometric information, and a wireless tag passed to all registrants and a personal ID input by a personal identification number are used as non-biometric authentication information. Authentication is performed using the following four authentication modes formed by combining information. First of all, since it is assumed that the administrator is in the vicinity in the daytime, even the registered person may not be able to pass, but it is unlikely that anyone other than the person will pass. The “daytime mode” is used, which allows the user to pass if the registered person is recognized in either of the non-biometric authentication information authentication in which the personal ID is input using the wireless tag. At night, authentication is performed in the normal safety priority mode. If you cannot pass, enter your personal ID using either the wireless tag or PIN, and identify the specific person specified by that with a high pass rate. The “night mode” in which authentication can be performed again by the “personal identification mode” described in the apparatus b of Example 4 is adopted. (FIG. 16) By using the daytime mode and the nighttime mode according to the time zone, it is possible to perform authentication with an emphasis on comfort at daytime and safety at nighttime.

  A block diagram of this security system is shown in FIG.

  Similarly to the first embodiment, the biological information receiving unit A includes three parts, that is, an image input unit 71, a face detection unit 72, and a feature extraction unit 73. The role of these parts is the same as that described in the first embodiment.

  The face image identification unit B includes a face matching unit 74 and a registered face image information storage unit 75. The registered face image information storage unit 75 is processing means for storing registered face image information necessary for collation and sending it to the face collation unit 74 when necessary. The face collation unit 74 collates and identifies the biometric feature information sent from the feature extraction unit 73 and the registered face image information sent from the registered face image storage unit 75 according to the threshold setting in the mode specified by the mode determination unit 76, and passes. It is a processing means which sends to the determination part 83 as identification result information. When authentication is performed in the personal identification mode, the face information storage unit 75 refers to the personal ID information of the person to be authenticated sent from the mode determination unit 76 and sends only the registered face image information of the specific individual to the face matching unit 74. Using the biometric feature information sent from the feature extracting unit 73, the face matching unit 74 performs personal identification.

  The non-biometric authentication information receiving unit F includes a wireless receiving unit 77 and a wireless information matching unit 78, a registered wireless information storage unit 79, a personal identification number input unit related to a personal ID input using a personal identification number. 80, a personal identification number collation unit 81, and a registered personal identification number storage unit 82. The wireless receiving unit 77 is a processing unit that senses a wireless tag owned by the person to be authenticated, acquires ID information, and sends the ID information to the wireless information matching unit 78 as personal ID information. The registered wireless information storage unit 79 is processing means for storing registered wireless information necessary for collation of the personal ID by the wireless tag and sending it to the wireless information collation unit 78 when necessary. The wireless information collating unit 78 collates and identifies the personal ID information sent from the wireless receiving unit 77 and the registered wireless information sent from the registered wireless information storage unit 79, and whether or not the person to be authenticated matches any of the registered persons. Is sent to the passage determining unit 83 as identification result information, and the personal ID selected as a result of the collation is sent to the mode determining unit 76.

  The personal identification number input unit 80 is a processing unit that acquires the personal identification number input by the authenticator as personal ID information and sends it to the personal identification number collation unit 81. The registered personal identification number storage unit 82 is a processing means for storing a registered personal identification number necessary for collating the personal ID with the personal identification number and sending it to the personal identification number collating unit 81 when necessary. The wireless information collation unit 81 collates and identifies the personal ID information transmitted from the personal identification number input unit 80 and the registered personal identification number transmitted from the registered personal identification number storage unit 82, and whether or not the person to be authenticated matches any of the registered persons. This is a processing means for sending the information as identification result information to the passage determination unit 83 and sending the personal ID selected as a result of the collation to the mode determination unit 76.

  The mode management unit C includes a mode determination unit 76 and a passage determination unit 83. The mode determination unit 76 instructs the face matching unit 74 to use the threshold setting used when identifying the threshold setting 1 if the time zone is daytime, and use the threshold setting 2 if the time zone is night. Information is sent to the passage determination unit 83, and either face image authentication or wireless tag authentication is performed in the safety priority mode in the daytime, and either face image authentication in the safety priority mode or authentication in the safety priority mode 2 is performed in the evening. Is a processing means for sending, as mode control information, information specifying that passage is permitted when the user succeeds. The passage determining unit 83 integrates the identification result information sent from the face matching unit 74, the wireless information matching unit 78, and the personal identification number matching unit 81, and finally permits passage according to the mode control information sent from the mode determining unit 76. It is a processing means to determine whether or not. When it is determined that the passage is not permitted in the safety priority mode at night, re-authentication request information for requesting the mode determination unit to perform the authentication in the personal identification mode is transmitted.

  The flow of processing when this security system is used will be described with reference to FIGS. If the time when the person to be authenticated visited is noon (step S71), if a face is detected in the input image (step S72), face image authentication is performed in the safety priority mode and the authentication result is “pass”. If so, the authentication ends (step S73). If the result is “pass” or if no face is detected in step S72 described above, the non-biometric authentication in the wireless tag mode is performed after sensing that the person to be authenticated possesses the wireless tag. Is performed (step S74). If the time zone is night, the face is first detected and face image authentication is performed in the safety priority mode (step S75). If the result is “pass”, the authentication ends there (step S76). If the result is “pass”, the personal ID is obtained from the person to be authenticated by the wireless tag or the personal identification number, and the face image is obtained again, and then the authentication in the personal identification mode is performed again (step S77).

  In the embodiment, in the daytime mode, the biometric information is first used for the determination, and then the biometric information is used for the final determination. However, the number of components of both pieces of information and their order are limited. Not. For example, in the night mode, the biometric information is checked after the wireless tag (non-biometric authentication information) is determined as in the individual identification mode used in steps S77 to S78 in FIG. This is to obtain a gain of limiting the search range of biometric information by acquiring non-biometric authentication information. For example, when the face detection range is wider than the wireless detection range, a daytime mode in which authentication is performed using a wireless tag may be used if the face image cannot be passed after authentication using a face image with an emphasis on speed. However, in order to perform authentication with higher security, it is better to first acquire a personal ID with a wireless tag and switch to a personal identification mode in which face image authentication is performed after narrowing the search range. In this way, the performance of the entire recognition can be controlled by the difference in order relation. Similar to the control described above, this control can be combined with other controls such as control by time or control by an administrator to perform mode management such as parameter adjustment.

  Each embodiment has been authenticated based on authentication using a face image so far, but any one of voice, fingerprint, palm print, iris, retina, etc. may be used as biometric information. Also, the type of non-biometric authentication information used in Examples 4 and 6 is not limited.

  The present invention is not limited to the above-described embodiment, and various modifications can be made within the technical scope thereof.

The figure which shows the basic composition of this invention. FIG. 3 is a diagram illustrating a configuration of an authentication mode according to the first embodiment. FIG. 3 is a diagram illustrating a processing procedure in the first embodiment. FIG. 10 is a diagram illustrating a configuration of an authentication mode in the second embodiment. FIG. 5 is a diagram illustrating a block configuration according to a second embodiment. FIG. 10 is a diagram illustrating a processing procedure in the second embodiment. FIG. 10 is a diagram illustrating a configuration of an authentication mode according to the third embodiment. FIG. 10 is a diagram illustrating a block configuration according to a third embodiment. The figure which shows the block configuration in Example 4 apparatus a. The figure which shows the process procedure in Example 4 apparatus a. The figure which shows the block configuration in Example 4 apparatus b. The figure which shows the process structure in Example 4 apparatus b. FIG. 10 is a diagram illustrating a configuration of an authentication mode according to a fifth embodiment. FIG. 10 is a diagram illustrating a block configuration according to a fifth embodiment. FIG. 10 is a diagram illustrating a processing configuration in a fifth embodiment. FIG. 10 is a diagram illustrating a configuration of an authentication mode according to a sixth embodiment. FIG. 10 is a diagram illustrating a block configuration according to a sixth embodiment. FIG. 10 is a diagram showing a processing configuration in a sixth embodiment. FIG. 10 is a diagram showing a processing configuration in a sixth embodiment. The figure which shows the relationship between a stranger exclusion rate and a person pass rate.

Explanation of symbols

DESCRIPTION OF SYMBOLS 41 Image input part 42 Face detection part 43 Feature extraction part 44 Collation part 45 Authentication information storage part 46 Mode determination part 49 External input device for to-be-authenticated persons

Claims (5)

A face image input unit for inputting the face image of the person to be authenticated,
A feature extraction unit for extracting feature information from the face image;
An authentication information storage unit for storing authentication face image registration information and identification information unique to the person to be authenticated in association with each other;
A collation unit that collates the feature information with the authentication-required registration information;
An identification information acquisition unit for acquiring identification information from the person to be authenticated; and
As a result of the collation performed by the collation unit, when the authentication registration information corresponding to the feature information cannot be specified, authentication face image registration information corresponding to the identification information acquired by the identification information acquisition unit is obtained. A re-collation control unit that extracts from the authentication information storage unit and causes the collation unit to perform collation again using the extracted authentication face image registration information and the feature information;
A face authentication device comprising:   The face authentication apparatus according to claim 1, wherein the identification information acquisition unit includes a wireless reception unit that receives the identification information from a wireless tag existing in a predetermined area.   The face authentication apparatus according to claim 1, wherein the identification information acquisition unit includes means for the certifier to input the identification information.   4. The face authentication apparatus according to claim 2, wherein the identification information is ID data unique to the person to be authenticated. A storing step of storing face image registration information for authentication of the person to be authenticated in association with ID data unique to the person to be authenticated;
A first input step for inputting the face image of the person to be authenticated;
A first collation step of collating the face image of the person to be authenticated input in the first input step with the face image registration information for authentication;
A second input step for inputting the ID data of the person to be authenticated and the face image when the first verification step is not successful;
The authentication face image registration information corresponding to the ID data input in the second input step is extracted from the authentication required face image registration information, and the registration information and the object to be input input in the second input step are extracted. A face verification method comprising: a second verification step of re-verifying the face image of the authenticator.

Images