Person authentication program, person authentication method, and person authentication device
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Filing Date
- 2023-08-30
- Publication Date
- 2026-05-11
Abstract
Description
Person authentication program, person authentication method, and person authentication device
[0001] The present invention relates to a person authentication program, a person authentication method, and a person authentication device.
[0002] In order to restrict the people passing through an entrance / exit, the face of the passerby is photographed from the front, an authentication process is performed, and the entrance / exit is opened or closed depending on the authentication result. In this case, in order to monitor unauthorized passersby who attempt to pass through the entrance / exit without authentication, a bird's-eye view photograph of the vicinity of the entrance / exit is taken from above, and a process of tracking the passerby is performed. Note that a person detection device is known that detects people from images captured by multiple cameras and associates and detects the same person from the images (see, for example, Patent Document 1).
[0003] Other known techniques include image tracking and techniques that automatically determine the timing of taking a photograph by recognizing a smile from a subject included in an image (see, for example, Patent Documents 2 to 5).
[0004] JP 2015-114917 A JP 2010-016878 A JP 2000-222576 A U.S. Patent Application Publication No. 2017 / 0178345 U.S. Patent Application Publication No. 2020 / 0074816
[0005] However, when performing authentication processing by photographing the faces of people, including passersby, from the front, there is a possibility that multiple people may accidentally appear in the authentication processing. In this case, even if a bird's-eye view photograph of the area around the entrance / exit is taken from above and people are tracked, it may not be possible to uniquely identify which person is the target of the authentication processing. For example, it is conceivable to impose a limit on the number of people who can be authenticated near the entrance / exit, but this could lead to congestion near the entrance / exit, which could reduce the convenience of the authentication processing.
[0006] It is also possible that if a specific pattern, such as a smile, is recognized in an overhead image, the person will be determined to be the target of authentication processing and tracking processing will be performed for that person. However, if the accuracy of this pattern identification is low, the target person for authentication processing may not be uniquely identified, and normal tracking processing may not be performed.
[0007] Therefore, in one aspect, an object is to provide a person authentication program, a person authentication method, and a person authentication device that uniquely identify a person who is a target of authentication processing.
[0008] In one embodiment, the person authentication program is a person authentication program to be executed by a computer connected to a first camera that captures an overall image of one or more persons in a monitored area and a second camera that captures an image of the head of the person when individually authenticating the person at the entrance to the monitored area, and when it recognizes that the person is in front of the second camera, it selects a specific head image that satisfies a condition from head images representing the head of the person that appear in the image captured by the first camera, compares the specific head image with the image captured by the second camera, and associates an identifier obtained as an authentication result with a specific person corresponding to the specific head image for which a similarity equal to or greater than a threshold similarity is calculated.
[0009] The person who is the subject of the authentication process can be uniquely identified.
[0010] 1 is an example of a person monitoring system. 2 is an example of a side view of the vicinity of an entrance in a monitored area. 3 is an example of the hardware configuration of a person authentication device. 4 is an example of the functional configuration of a person authentication device. (a) is an example of authentication information. (b) is another example of authentication information. (a) is an example of a wide-area image in which two people appear. (b) is an example of a wide-area image in which person detection and head detection have been performed. (b) is an example of second head information. 5 is a flowchart showing an example of processing performed by the person authentication device. (b) is another example of second head information. (c) is an example of generation of first head information. (d) is an example of selection of an optimal head image. (a) is an example of calculation of similarity. (b) is an example of matching of a person's ID. (c) is an example of a wide-area image in which one person appears. (a) is an example of a wide-area image in which the distance between the person and the authentication camera is equal to or greater than a threshold distance. (b) is an example of a wide-area image explaining the respective distances between two people and the authentication camera.
[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
[0012] 1, the person monitoring system ST includes a monitoring camera 10, an authentication camera 20, an authentication information reader 21, and a person authentication device 100. The monitoring camera 10 is an example of a first camera, and the authentication camera 20 is an example of a second camera. The person authentication device 100 is an example of a computer. The person monitoring system ST may include multiple monitoring cameras 11, 12, 13, 14, and 15.
[0013] The surveillance cameras 10, 11, 12, 13, 14, and 15, the authentication camera 20, and the authentication information reader 21 are connected to the person authentication device 100 via a communication network NW. The communication network NW includes at least one of a LAN (Local Area Network), a WAN (Wide Area Network), and the Internet.
[0014] An input device 710 and a display device 720 are connected to the person authentication device 100. Examples of the input device 710 include a keyboard, a mouse, and a touch panel. Examples of the display device 720 include a liquid crystal display. The input device 710 is operated, for example, by an administrator P0 of the person monitoring system ST. Based on an operation on the input device 710, the person authentication device 100 displays, for example, a wide-area image captured by the surveillance camera 10 (hereinafter simply referred to as a wide-area image) on the display device 720. Furthermore, based on the wide-area image captured by at least one of the surveillance cameras 10, 11, 12, 13, 14, and 15, the person authentication device 100 associates and tracks the same person P1 appearing in the wide-area image.
[0015] The surveillance cameras 10, 11, 12, 13, 14, and 15 each capture a bird's-eye view of the monitored area AR in color. For example, if the monitored area AR is an office, the surveillance cameras 10, 11, 12, 13, 14, and 15 are installed above the heads of the persons P1 and P2, such as on the ceiling. As a result, for example, the surveillance camera 10 captures an image of the entire captured area, including the entire bodies of the persons P1 and P2 in the monitored area AR, the authentication camera 20, the authentication information reader 21, and the like. Note that the monitored area AR is not limited to an office, and may also be a medical facility such as a hospital or clinic, a public facility such as an airport or train station, or a commercial facility such as a shopping mall.
[0016] Furthermore, the surveillance camera 10 is installed so as to capture an image of person P1 positioned in front of the authentication camera 20 and the authentication information reader 21. As a result, person P1 appears partially or entirely in the wide-area image captured by the surveillance camera 10. In this case, the surveillance camera 10 is installed so that the authentication camera 20 or the authentication information reader 21 does not overlap with the head of person P1 in the wide-area image. As a result, the face of person P1 appears in the wide-area image captured by the surveillance camera 10 without being hidden by the authentication camera 20 or the authentication information reader 21.
[0017] The authentication camera 20 captures, for example, a color image of the head of person P1 from the front of person P1. Specifically, as shown in Fig. 2, the authentication camera 20 is installed above a pedestal 22 via a tripod. The imaging range of the authentication camera 20 is adjusted so as to capture an image of the head Ph of person P1 from the front. Note that Fig. 2 corresponds to the predetermined area ARy shown in Fig. 1. The tripod may be extended and the pedestal 22 may be removed, or the tripod may be removed from the authentication camera 20 and the authentication camera 20 may be installed directly on the top surface of the pedestal 22.
[0018] Meanwhile, as shown in FIG. 1 , the authentication information reader 21 individually authenticates, for example, person P1 at the entrance ARx of the monitoring area AR. As shown in FIG. 2 , the authentication information reader 21 is installed on the top surface of a base 22. The authentication information reader 21 is also installed in front of the authentication camera 20. The authentication information reader 21 can authenticate person P1 by reading biometric information related to biometric authentication and communicating with the person authentication device 100. The authentication information reader 21 can also authenticate person P1 by reading card information (e.g., a personal identification number) recorded on an ID (identifier) card carried by person P1 and communicating with the person authentication device 100. The ID card is an example of a person identification medium. In this way, the authentication information reader 21 individually authenticates person P1, and the authentication camera 20 supplementarily captures an image of person P1's head Ph during the authentication process by the authentication information reader 21.
[0019] The biometric authentication described above may be fingerprint authentication, vein authentication, or iris authentication. Furthermore, in this embodiment, the authentication of person P1 is described using both the authentication camera 20 and the authentication information reader 21. On the other hand, if a face authentication camera is used as the authentication camera 20, the authentication information reader 21 may be excluded from the person monitoring system ST. The face authentication camera can authenticate person P1 using a face image (i.e., an image of the face) by communicating with the person authentication device 100.
[0020] In this way, when person P1 enters the monitoring area AR, he or she has the authentication information reader 21 read the biometric information of person P1 himself or herself or the card information of the ID card carried by person P1. Because the authentication information reader 21 is installed in front of the authentication camera 20, when the authentication information reader 21 authenticates person P1, the authentication camera 20 captures an image of the head Ph of person P1. The biometric information or card information is transmitted from the authentication information reader 21 to the person authentication device 100. An image of the head of person P1 captured at the time of authentication, including the head Ph of person P1 (hereinafter referred to as an authentication head image), is transmitted from the authentication camera 20 to the person authentication device 100.
[0021] The person authentication device 100 authenticates person P1 based on biometric information or card information, and stores the authentication result and an authenticated head image of person P1 in association with each other as first head information. As will be described in detail later, when the person authentication device 100 receives the authenticated head image of person P1, it estimates the direction of head Ph based on the authenticated head image. Specifically, the person authentication device 100 calculates a head direction vector based on the authenticated head image. The head direction vector represents, for example, the facial orientation of person P1. The person authentication device 100 also stores this head direction vector as first head information in association with the authentication result and the like.
[0022] Meanwhile, the surveillance camera 10 captures an image of the authentication information reader 21 authenticating person P1. The wide-area image of the surveillance camera 10 is periodically (e.g., every few seconds) transmitted from the surveillance camera 10 to the person authentication device 100. As shown in Fig. 1, if the imaging range of the surveillance camera 10 includes not only person P1 but also person P2, the entire bodies of both person P1 and person P2 will appear in the imaging range of the surveillance camera 10.
[0023] As will be described in detail later, when the person authentication device 100 receives a wide-area image from the surveillance camera 10, it detects the entire bodies of persons P1 and P2 appearing in the wide-area image and assigns different temporary identification information (hereinafter referred to as temporary IDs) to each of persons P1 and P2. Furthermore, the person authentication device 100 detects the heads of persons P1 and P2 from the full-body images and estimates the head directions based on head images representing the detected heads (hereinafter referred to as detected head images). In other words, the person authentication device 100 calculates a head direction vector.
[0024] When the person authentication device 100 calculates the head direction vector, it associates the temporary ID, the head direction vector, the detected head image, and the acquisition date and time of the wide-angle image and stores them as second head information. After storing the second head information, the person authentication device 100 identifies the same person based on the first head information and the second head information. That is, the person authentication device 100 identifies person P1, who is the person to be authenticated and who is in front of the authentication camera 20, from among persons P1 and P2 appearing in the wide-angle image of the surveillance camera 10. When the person authentication device 100 identifies the same person, it replaces the temporary ID assigned to person P1 with a real ID that identifies person P1.
[0025] In this way, the person authentication device 100 can accurately associate a real ID with the person P1 to whom a temporary ID has been assigned, without detecting a specific pattern such as a smile, and can subsequently track the person P1 as he or she moves within the monitoring area AR. As a result, even if the person P1 commits an illegal act within the monitoring area AR, the person authentication device 100 can uniquely identify the person P1. Furthermore, the person authentication device 100 can uniquely grasp the movement of the person P1 within the monitoring area AR.
[0026] Next, the hardware configuration of the person authentication device 100 will be described with reference to FIG.
[0027] The person authentication device 100 includes a CPU (Central Processing Unit) 100A as a processor, and a RAM (Random Access Memory) 100B and a ROM (Read Only Memory) 100C as memories. The RAM 100B includes a DRAM (Dynamic RAM) and an SRAM (Static RAM). The SRAM may be included in the CPU 100A. The person authentication device 100 also includes a network I / F (Interface) 100D and an HDD (Hard Disk Drive) 100E. An SSD (Solid State Drive) may be used instead of the HDD (Hard Disk Drive) 100E.
[0028] The person authentication device 100 may include at least one of an input I / F 100F, an output I / F 100G, an input / output I / F 100H, and a drive device 100I, as necessary. The CPU 100A to the drive device 100I are connected to each other by an internal bus 100J. In other words, the person authentication device 100 can be realized by a computer.
[0029] An input device 710 is connected to the input I / F 100F. A display device 720 is connected to the output I / F 100G. A semiconductor memory 730 is connected to the input / output I / F 100H. Examples of the semiconductor memory 730 include a USB (Universal Serial Bus) memory and a flash memory. The input / output I / F 100H reads the person authentication program stored in the semiconductor memory 730. The input I / F 100F and the input / output I / F 100H each include, for example, a USB port. The output I / F 100G includes, for example, a display port.
[0030] A portable recording medium 740 is inserted into the drive device 100I. The portable recording medium 740 may be a removable disk such as a CD (Compact Disc)-ROM or a DVD (Digital Versatile Disc). The drive device 100I reads the person authentication program recorded on the portable recording medium 740. The network I / F 100D includes, for example, a LAN port and a communication circuit. The communication circuit includes either a wired communication circuit or a wireless communication circuit, or both. The network I / F 100D is connected to a communication network NW.
[0031] A person authentication program stored in at least one of ROM 100C, HDD 100E, and semiconductor memory 730 is temporarily stored in RAM 100B by CPU 100A. A person authentication program recorded on portable recording medium 740 is temporarily stored in RAM 100B by CPU 100A. By executing the stored person authentication program, CPU 100A realizes various functions described below and executes a person authentication method including various processes described below. The person authentication program may be one that corresponds to the flowchart described below.
[0032] The functional configuration of the person authentication device 100 will be described with reference to Fig. 4 to Fig. 7. Note that Fig. 4 shows the main functions of the person authentication device 100.
[0033] As shown in Fig. 4, the person authentication device 100 includes a storage unit 110, a processing unit 120, an input unit 130, an output unit 140, and a communication unit 150. The storage unit 110 can be realized by either or both of the RAM 100B and the HDD 100E described above. The processing unit 120 can be realized by the CPU 100A described above. The input unit 130 can be realized by the input I / F 100F described above. The output unit 140 can be realized by the output I / F 100G described above. The communication unit 150 can be realized by the network I / F 100D described above.
[0034] The memory unit 110, processing unit 120, input unit 130, output unit 140, and communication unit 150 are connected to one another. The memory unit 110 includes an authentication information DB (Data Base) 111 and a head information DB 112. The processing unit 120 includes a head information generation unit 121 and a person identification unit 122. The person identification unit 122 is an example of a selection means and an association means.
[0035] The authentication information DB 111 stores in advance authentication information required for authenticating multiple persons including persons P1 and P2. As shown in FIG. 5A, the authentication information includes multiple items such as an authentication ID, an authentication type, biometric information, card information, and a person ID. An identifier that identifies the authentication information is registered in the authentication ID item. An authentication method is registered in the authentication type item. Examples of authentication methods include biometric authentication and card authentication.
[0036] In the biometric information item, biometric information that uniquely identifies each of multiple persons is registered. When the authentication method is biometric authentication, biometric information is registered in the biometric information item. In the card information item, card information recorded on an ID card carried by each of multiple persons is registered. When the authentication method is card authentication, card information is registered in the card information item. In the person ID item, identifiers that identify multiple persons are registered. For example, if biometric information "##%#%#%" is registered in the biometric information item as the biometric information of person P1, the person ID "tanaka" is uniquely identified as the person ID of person P1 based on this biometric information "##%#%#%".
[0037] When a face recognition camera is used as the authentication camera 20, the authentication information may include multiple items such as an authentication ID, a face image, and a person ID, as shown in Fig. 5(b). For example, if a face image with a pentagonal outline is registered as the face image of person P1 in the face image item, the person ID "tanaka" is uniquely identified as the person ID of person P1 based on this face image. For example, hair color may be used for authentication based on the face image.
[0038] The head information generation unit 121 generates the first head information and the second head information described above. When the head information generation unit 121 generates the first head information, it stores the first head information. When the head information generation unit 121 generates the second head information, it stores the second head information in the head information DB 112.
[0039] More specifically, the head information generation unit 121 periodically acquires wide-area images from the surveillance camera 10. For example, as shown in FIG. 6( a), when a wide-area image is acquired, the head information generation unit 121 performs person detection on all persons P1 and P2 appearing in the wide-area image based on a known person detection technology using OpenCV, as shown in FIG. 6( b). After performing person detection, the head information generation unit 121 assigns a temporary ID to each of the persons P1 and P2. For example, the head information generation unit 121 assigns a temporary ID of "1" to the person P1 and a temporary ID of "2" to the person P2.
[0040] After assigning the temporary IDs, the head information generation unit 121 performs face detection (or face recognition) on each of the persons P1 and P2 based on a known face detection technology (or face recognition technology) using OpenCV, and acquires detected head images. After acquiring the detected head images, the head information generation unit 121 calculates head direction vectors for each of the persons P1 and P2 based on the detected head images and a known head direction estimation technology (or face orientation estimation technology) such as HOPE-Net or FaceRig.
[0041] This head direction estimation technology includes a process of detecting the coordinates of six landmarks, such as the corners of the eyes, the tip of the nose, and the corners of the mouth, from the detected head image, and calculating the virtual position of the camera based on the detected coordinates to estimate the head direction. The head direction (or face orientation) of each of the persons P1 and P2 is estimated from the head direction vector calculated by the head information generation unit 121.
[0042] In this way, after calculating the head direction vector, the head information generation unit 121 stores second head information in which the tentative ID, the head direction vector, the detected head image, and the acquisition date and time of the wide-area image are associated with each other in the head information DB 112. As a result, the head information DB 112 stores the second head information of each of the persons P1 and P2, as shown in Fig. 7 .
[0043] The second head information includes multiple items such as a temporary ID, a head direction vector, a head image, and acquisition date and time. A temporary ID is registered in the temporary ID field. A head direction vector is registered in the head direction vector field. The head direction vector is expressed, for example, by a roll angle "r4", a pitch angle "p4", and a yaw angle "y4". A detected head image is registered in the head image field. An acquisition date and time is registered in the acquisition date and time field. The acquisition date and time is expressed, for example, in the YYYYMMDDHHMMSS format. Note that the head information generation unit 121 generates the first head information basically in the same way as the second head information except for the acquisition date and time, but details of generating the first head information will be described later.
[0044] The person identification unit 122 identifies the same person based on the first head information and the second head information. More specifically, the person identification unit 122 determines whether each detected head image from among the detected head images included in the second head information satisfies a predetermined condition, and selects a specific detected head image that satisfies the predetermined condition. The predetermined condition is, for example, a condition between the head direction vector included in the first head information and the head direction vector included in the second head information. In other words, the predetermined condition is a condition related to the angle of the head direction. When the head direction vectors are most similar to each other, that is, when the angles between the head directions are most similar to each other, the person identification unit 122 determines that the predetermined condition is satisfied.
[0045] In this case, the person identification unit 122 selects, for each temporary ID, the detected head image associated with the head direction vector that is most similar to the head direction vector included in the first head information as a specific optimal head image. Then, the person identification unit 122 compares the optimal head image for each temporary ID with the authenticated head image captured by the authentication camera 20, and associates the person ID obtained as the authentication result with either of the specific persons P1 and P2 corresponding to the optimal head image whose similarity is calculated to be equal to or greater than the threshold similarity.
[0046] As a result, even if persons P1 and P2 appear in the wide-area image, the person identification unit 122 can accurately identify, for example, person P1 as the target person for authentication processing. Thereafter, the person identification unit 122 can track person P1 based on the wide-area images of the multiple surveillance cameras 10, 11, 12, 13, 14, and 15. Note that the person identification unit 122 may select, as specific optimal head images for each temporary ID, detected head images associated with the head direction vector that is most similar and the head direction vector that is second most similar to the head direction vector included in the first head information.
[0047] Next, an example of the operation of the person authentication device 100 will be described with reference to FIGS.
[0048] First, as shown in FIG. 8 , the head information generation unit 121 acquires a wide-angle image from the surveillance camera 10 (step S1). After acquiring the wide-angle image, the head information generation unit 121 performs person detection (step S2) and stores the second head information in the head information DB 112 (step S3). After storing the second head information, the head information generation unit 121 determines whether or not person P1 has been authenticated (step S4). For example, unless person P1 has the authentication information reader 21 read his or her own biometric information or card information, the head information generation unit 121 determines that person P1 has not been authenticated (step S4: NO). As a result, the head information generation unit 121 repeats the processes from steps S1 to S3.
[0049] 9, the head information DB 112 stores multiple pieces of second head information while adding them every few seconds. The head information generation unit 121 may store the second head information in the head information DB 112 until a predetermined storage period (for example, a few minutes) and delete the second head information after the storage period has expired from the head information DB 112. This makes it possible to prevent a decrease in the storage capacity of the head information DB 112.
[0050] Here, for example, when person P1 performs an authentication action by having the authentication information reader 21 read person P1's own biometric information, the head information generation unit 121 determines that person P1 has been authenticated (step S4: YES). If person P1 has been authenticated, the head information generation unit 121 acquires person P1's biometric information from the authentication information reader 21. As a result, the head information generation unit 121 can identify person P1's person ID "tanaka" as the authentication result, as shown in the upper part of FIG. 10, based on the biometric information and the authentication information stored in the authentication information DB 111 (see FIG. 5(a)). After identifying the authentication result, the head information generation unit 121 determines whether two or more people appear in the most recent wide-area image (step S5).
[0051] For example, if persons P1 and P2 appear in the wide-angle image (see FIG. 6A), and two or more persons appear in the wide-angle image (step S5: YES), the head information generation unit 121 acquires an authentication head image from the authentication camera 20 (step S6). That is, person P1, who performed the authentication action on the authentication information reader 21, is standing facing forward, less than 1 m (meter) in front of the authentication camera 20. Therefore, as shown in the middle part of FIG. 10, the head information generation unit 121 can accurately acquire the authentication head image captured by the authentication camera 20.
[0052] When the authentication head image is acquired, the head information generation unit 121 generates and stores first head information (step S7). That is, as shown in the lower part of Fig. 10, the head information generation unit 121 generates and stores first head information that associates the person ID "tanaka" identified as the authentication result with the authentication head image and the head direction vector calculated based on the authentication head image.
[0053] The first head information includes multiple items such as a person ID, a head direction vector, and a head image. The person ID identified as the authentication result is registered in the person ID item. The head direction vector calculated based on the authenticated head image is registered in the head direction vector item. The authenticated head image is registered in the head image item. In this way, the first head information differs from the second head information in that it does not include the acquisition date and time.
[0054] When the head information generation unit 121 generates and stores the first head information, the person identification unit 122 selects an optimal head image (step S8). More specifically, as shown in FIG. 11 , the person identification unit 122 calculates the vector difference between the head direction vector included in the first head information and the head direction vector included in the second head information for each angle. This calculates the vector difference for each roll angle, pitch angle, and yaw angle. After calculating the vector difference, the person identification unit 122 squares the vector difference for each angle to calculate its sum, and selects, for each temporary ID, a head image associated with the head direction vector that minimizes the sum as a specific optimal head image. That is, the person identification unit 122 identifies, based on the least squares method, a head direction vector from the second head information that is most similar to the head direction vector included in the first head information. Then, the person identification unit 122 selects, for each temporary ID, a detected head image associated with the identified head direction vector as a specific optimal head image.
[0055] In this embodiment, as shown in Fig. 11, the person identification unit 122 selects the detected head image of the second head information including the temporary ID "1" and the head direction vector "r4, p4, y4" as the specific optimum head image. Also, the person identification unit 122 selects the detected head image of the second head information including the temporary ID "2" and the head direction vector "r5, p5, y5" as the specific optimum head image.
[0056] After selecting the optimal head image, the person identification unit 122 calculates the similarity (step S9). More specifically, as shown in Fig. 12(a), the person identification unit 122 calculates the image feature of the authentication head image 50 included in the first head information and the image feature of each of the two selected optimal head images 51 and 52, and calculates the similarity between the authentication head image 50 and the optimal head images 51 and 52 based on the image feature. For example, the person identification unit 122 can calculate the image feature based on the hair color and facial contour of the heads of the authentication head image 50 and the optimal head images 51 and 52, and calculate the similarity.
[0057] In this embodiment, the person identification unit 122 calculates a similarity of 98% between the authenticated head image 50 included in the first head information and the optimal head image 51 associated with the temporary ID "1." The person identification unit 122 also calculates a similarity of 12% between the authenticated head image 50 included in the first head information and the optimal head image 52 associated with the temporary ID "2." In this way, the number of optimal head images to be used for calculating similarities is reduced in advance by the head direction vector, so the person identification unit 122 can complete the calculation of similarities in a shorter time than when calculating all similarities.
[0058] After calculating the similarity, the person identification unit 122 determines whether the similarity is equal to or greater than a threshold (step S10). More specifically, the person identification unit 122 determines whether the similarity is equal to or greater than a threshold similarity for each optimal head image. The threshold similarity can be set appropriately in advance, and in this embodiment, for example, 90% is set as the threshold similarity. If all of the similarities are less than the threshold (step S10: NO), the head information generation unit 121 returns to the processing of step S1.
[0059] On the other hand, if any of the similarities is equal to or greater than the threshold (step S10: YES), the person identification unit 122 associates the person ID (step S11) and terminates the process. More specifically, as shown in FIG. 12(b), the person identification unit 122 associates the person ID with the optimal head image 51 by replacing the temporary ID "1" associated with the optimal head image 51 for which the similarity equal to or greater than the threshold is calculated with the personal ID "tanaka". Furthermore, the person identification unit 122 associates the person ID with the optimal head image 51 by replacing the temporary ID "1" associated with person P1 appearing in the wide-area image with the personal ID "tanaka". In this way, the person identification unit 122 can identify person P1 appearing in the wide-area image and person P1 who is the target of the authentication process.
[0060] In the process of step S5, if two or more people do not appear in the wide-area image (step S5: NO), the person identification unit 122 executes the process of step S10 and ends the process. For example, as shown in FIG. 13 , when person P1 appears alone in the wide-area image, person P2 does not appear in the wide-area image, and therefore the person identification unit 122 can uniquely identify person P1 as the person to whom the personal ID is to be associated. In this way, when two or more people do not appear in the wide-area image, the head information generation unit 121 and the person identification unit 122 can avoid the processes of steps S6 to S10. In other words, when two or more people do not appear in the wide-area image, the person identification unit 122 can associate the personal ID in a short time with a reduced processing load.
[0061] 14A, when the authentication camera 20 is a face recognition camera, if the distance between the person P1 and the authentication camera 20 is, for example, 1 meter or more, the authentication camera 20 may not be able to accurately capture an authentication head image of the person P1. When the authentication information reader 21 is used, the person P1 approaches the authentication information reader 21, and therefore the distance between the person P1 and the authentication camera 20 is likely to be less than 1 meter. However, when a face recognition camera is used, the distance between the person P1 and the authentication camera 20 may be 1 meter or more, and the authentication head image of the person P1 may not be able to be accurately captured.
[0062] For this reason, when a face recognition camera is used, if two or more people appear in the wide-area image in the processing of step S5, immediately after that processing, as shown in FIG. 14(b), the head information generation unit 121 determines whether or not the distances between the people P1, P2 and the authentication camera 20 are each less than a threshold distance. Based on multiple objects included in the wide-area image, the head information generation unit 121 can determine whether or not the distances between the people P1, P2 and the authentication camera 20 are each less than a threshold distance. The threshold distance may be a few meters, such as 1 meter, or may be several tens of centimeters.
[0063] If both the distance between person P1 and the authentication camera 20 and the distance between person P2 and the authentication camera 20 are equal to or greater than the threshold distance, the head information generation unit 121 determines that the person to be authenticated is not present, and the process returns to step S1. This allows the head information generation unit 121 and the person identification unit 122 to avoid the processes of steps S6 to S10.
[0064] On the other hand, if the distance between person P1 and the authentication camera 20 is less than the threshold distance and the distance between person P2 and the authentication camera 20 is equal to or greater than the threshold distance, the person identification unit 122 may uniquely identify person P1 as the target person for authentication processing. Even in this case, the head information generation unit 121 and the person identification unit 122 can avoid the processing of steps S6 to S10.
[0065] If both the distance between person P1 and the authentication camera 20 and the distance between person P2 and the authentication camera 20 are less than the threshold distance, the head information generation unit 121 and the person identification unit 122 perform the processing from steps S6 to S9 and then execute the processing of step S10.
[0066] Although the preferred embodiments of the present invention have been described in detail above, the present invention is not limited to the specific embodiments, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims.
[0067] For example, the person identification unit 122 determines whether to select a specific optimal head image that satisfies the above-mentioned predetermined conditions based on the degree of association between the facility in the monitored area AR and the type of clothing commonly used at that facility. If the degree of association is higher than a threshold degree of association, the person identification unit 122 avoids selecting the specific optimal head image that satisfies the predetermined conditions. For example, if a person wearing a white coat commonly used in a facility such as a hospital appears in the wide-area image, it is assumed that the person is unlikely to commit fraud. Therefore, by excluding such people from tracking targets, the person identification unit 122 can limit tracking targets to a limited number of people, thereby reducing the processing load and identifying people.
[0068] ST Person monitoring system AR Monitoring target area P1, P2 Person 10, 11, 12, 13, 14, 15 Monitoring camera 20 Authentication camera 21 Authentication information reader 100 Person authentication device 110 Storage unit 111 Authentication information DB 112 Head information DB 120 Processing unit 121 Head information generation unit 122 Person identification unit
Claims
1. A person authentication program to be executed on a computer connected to a first camera that captures an overall image of one or more people in a monitored area, and a second camera that captures an image of the head of a person when individually authenticating that person at the entrance to the monitored area, When the second camera recognizes that the person is in front of it, it selects a specific head image that satisfies the conditions from among the head images representing the person's head that appear in the image captured by the first camera. The specific head image is compared with the image captured by the second camera, and the identifier obtained as an authentication result is associated with the specific person corresponding to the specific head image for which a similarity of a threshold similarity or higher has been calculated. A person authentication program for causing the computer to perform the processing.
2. The process of selecting a specific head image involves, based on the image captured by the first camera, recognizing that the person is in front of the second camera, selecting a specific head image from among the head images representing the person's head that appear in the image captured by the first camera, whose head direction is similar to that of the head image appearing in the image captured by the second camera. The person authentication program according to feature 1.
3. The process of selecting a specific head image involves, based on the image captured by the first camera, detecting the head of the person appearing in the image captured by the first camera when the second camera recognizes that the person is in front of it, and storing information in a database that associates the head image representing the detected head with the head direction estimated based on the head image and a temporary identifier that identifies the person corresponding to the head image at each unit time, until the information's retention period expires, and deleting the information from the database after the retention period has expired. A person authentication program according to feature 1 or 2.
4. The process of selecting a specific head image avoids selecting a specific head image that satisfies the conditions if, based on the image captured by the first camera, it recognizes that the person in front of the second camera is alone. A person authentication program according to feature 1 or 2.
5. The process of selecting a specific head image avoids selecting a specific head image that satisfies the conditions if, based on the image captured by the first camera, it is recognized that the person is located at a distance greater than or equal to a threshold distance from the front of the second camera. A person authentication program according to feature 1 or 2.
6. The process of selecting a specific head image involves determining whether or not to select a specific head image that satisfies the conditions based on the degree of association between the facility in the monitored area and the type of clothing commonly used at the facility, and if the degree of association is higher than the threshold degree of association, the selection of a specific head image that satisfies the conditions is avoided. A person authentication program according to feature 1 or 2.
7. The second camera is a facial recognition camera that authenticates the face of the person. A person authentication program according to feature 1 or 2.
8. The second camera is installed within the imaging range of the first camera. A person authentication program according to feature 1 or 2.
9. A person authentication method performed by a computer connected to a first camera that captures an overall image of one or more people in a monitored area, and a second camera that captures an image of the head of a person when individually authenticating that person at the entrance to the monitored area, When the second camera recognizes that the person is in front of it, it selects a specific head image that satisfies the conditions from among the head images representing the person's head that appear in the image captured by the first camera. The specific head image is compared with the image captured by the second camera, and the identifier obtained as an authentication result is associated with the specific person corresponding to the specific head image for which a similarity of a threshold similarity or higher has been calculated. A person authentication method in which the computer performs the processing.
10. A person authentication device connected to a first camera that captures an overall image of one or more people in a monitored area, and a second camera that captures an image of a person's head when individually authenticating the person at the entrance to the monitored area, When the second camera recognizes that the person is in front of it, a selection means selects a specific head image that satisfies certain conditions from among the head images representing the person's head that appear in the image captured by the first camera. A matching means that compares the specific head image with the image captured by the second camera and associates the identifier obtained as an authentication result with a specific person corresponding to the specific head image for which a similarity of a threshold similarity or higher has been calculated. A person authentication device that includes [this].