WO2022244357A1 - 生体の認証システムおよび認証方法 - Google Patents
生体の認証システムおよび認証方法 Download PDFInfo
- Publication number
- WO2022244357A1 WO2022244357A1 PCT/JP2022/006798 JP2022006798W WO2022244357A1 WO 2022244357 A1 WO2022244357 A1 WO 2022244357A1 JP 2022006798 W JP2022006798 W JP 2022006798W WO 2022244357 A1 WO2022244357 A1 WO 2022244357A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- feature amount
- authentication
- period
- face
- feature
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 116
- 238000012545 processing Methods 0.000 claims abstract description 123
- 238000003860 storage Methods 0.000 claims abstract description 24
- 238000003384 imaging method Methods 0.000 claims description 26
- 238000000605 extraction Methods 0.000 claims description 21
- 230000004927 fusion Effects 0.000 claims description 12
- 230000001144 postural effect Effects 0.000 abstract 1
- 230000001815 facial effect Effects 0.000 description 75
- 230000008569 process Effects 0.000 description 54
- 238000010586 diagram Methods 0.000 description 30
- 239000000872 buffer Substances 0.000 description 27
- 238000005516 engineering process Methods 0.000 description 25
- 230000036544 posture Effects 0.000 description 21
- 210000003462 vein Anatomy 0.000 description 21
- 238000001514 detection method Methods 0.000 description 17
- 230000000694 effects Effects 0.000 description 14
- 230000007704 transition Effects 0.000 description 14
- 230000008901 benefit Effects 0.000 description 13
- 238000004891 communication Methods 0.000 description 8
- 239000013598 vector Substances 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 7
- 210000004204 blood vessel Anatomy 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000010606 normalization Methods 0.000 description 5
- XUMBMVFBXHLACL-UHFFFAOYSA-N Melanin Chemical compound O=C1C(=O)C(C2=CNC3=C(C(C(=O)C4=C32)=O)C)=C2C4=CNC2=C1C XUMBMVFBXHLACL-UHFFFAOYSA-N 0.000 description 4
- 230000006866 deterioration Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 210000000554 iris Anatomy 0.000 description 4
- 238000004904 shortening Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 230000003139 buffering effect Effects 0.000 description 3
- 210000000887 face Anatomy 0.000 description 3
- 238000010187 selection method Methods 0.000 description 3
- 210000003491 skin Anatomy 0.000 description 3
- 244000060701 Kaempferia pandurata Species 0.000 description 2
- 235000016390 Uvaria chamae Nutrition 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000010349 pulsation Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 208000035473 Communicable disease Diseases 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000004397 blinking Effects 0.000 description 1
- 230000017531 blood circulation Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000593 degrading effect Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000005069 ears Anatomy 0.000 description 1
- 210000002615 epidermis Anatomy 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 210000000744 eyelid Anatomy 0.000 description 1
- 230000008921 facial expression Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 210000001145 finger joint Anatomy 0.000 description 1
- 210000003128 head Anatomy 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012905 input function Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
- G06F21/32—User authentication using biometric data, e.g. fingerprints, iris scans or voiceprints
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/117—Identification of persons
- A61B5/1171—Identification of persons based on the shapes or appearances of their bodies or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/14—Vascular patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/70—Multimodal biometrics, e.g. combining information from different biometric modalities
Definitions
- the present invention relates to an authentication system and an authentication method for authenticating an individual using biometric information.
- biometric authentication technology has been used as a means of reducing the risk of information leakage and unauthorized use of mobile terminals such as smartphones and notebook PCs.
- mobile terminals used in remote environments have a high risk of being illegally used by others. Therefore, when accessing a terminal or information system, it is required to perform personal authentication each time. However, it is troublesome to enter a password every time, and there is a risk of forgetting or leaking the password.
- Cashless payments are becoming popular at retail stores such as convenience stores and restaurants.
- Cashless payment is highly convenient because there is no need to pay cash on the spot, and it is also highly advantageous for stores to introduce it because it can promote customer purchases by linking with various point services. If biometric authentication is used for such cashless payment, there is no need to carry a card or the like, and since the identity of the person can be confirmed with certainty, a convenient and effective service can be provided.
- biometric authentication has many advantages in preventing unauthorized access and realizing cashless payments
- the need for additional dedicated biometric authentication devices increases the introduction cost, which hinders its widespread use.
- biometric authentication can be performed using biometric images captured by general-purpose cameras installed in smartphones, notebook PCs, etc.
- the barriers to introducing biometric authentication can be lowered.
- the authentication operation can be performed without contact, the risk of the spread of infectious diseases, which is a recent social problem, will be reduced, so it is thought that it can be introduced and used with peace of mind.
- biometric authentication the user's own biometrics such as fingers, hands, and face are held over the authentication terminal, and the identity of the registrant is confirmed after matching with pre-registered biometric information. Login and payment are performed only when the registered user is authenticated.
- biometric authentication based on features of the inside of a living body such as finger veins is known as one capable of realizing highly accurate authentication.
- Finger vein authentication uses a complicated blood vessel pattern inside the finger, so it achieves excellent authentication accuracy, and is more difficult to forge and falsify than fingerprint authentication, so it can achieve a high level of security.
- Biometric authentication using a general-purpose camera is more difficult to photograph under suitable conditions than biometric authentication using a dedicated sensor that specializes in photographing a living body, and authentication accuracy tends to decrease due to deterioration in image quality.
- multimodal biometric authentication technology that improves authentication accuracy by using multiple pieces of biometric information is effective. Basically, it is relatively easy to take images at the same time, and it is possible to combine a plurality of biometric features with low correlation between biometric features or imaging conditions, and play complementary roles in each biometric feature. It is possible to effectively improve authentication accuracy.
- the former tends to take a long time to shoot, while the latter poses a challenge in that it is difficult to hold the body over at the same time.
- the face may be tilted or the face may be blocked by the held finger.
- Patent Document 1 is a multimodal authentication technology that simultaneously performs face and fingerprint matching.
- Patent Document 2 discloses a technique for measuring the flatness of an image of a face while using the face and fingerprints for authentication.
- Japanese Patent Laid-Open No. 2002-200001 has a first phase in which face matching is performed using face data, and a second phase in which finger matching and face photographing are performed.
- a technique for detecting the face orientation (up, down, left, or right) of a face image while photographing a face is disclosed.
- face image photography in the second phase is used not for face authentication but for gesture determination, and there is no mention of technology for solving problems related to accuracy improvement and speeding up of multimodal authentication.
- Patent Document 2 when the finger is placed on the fingerprint sensor, the illumination for photography is turned on toward the face, and the face is photographed by the camera. A technology has been disclosed in which each feature amount for face authentication and each feature amount for fingerprint authentication are put into the same category, and individual authentication is performed using a minimum distance identification method or the like. Although Patent Document 2 discloses the viewpoint of performing highly accurate multimodal biometric authentication using a face and fingerprints, in addition to the need for a fingerprint sensor, accuracy can be improved by suppressing posture fluctuations. There is no mention of planning techniques.
- the above-mentioned problems are not limited to multimodal biometric authentication of the face and fingers, but also include iris, auricle, facial vein, subconjunctival blood vessel, palm vein, back vein, palm print, joint print inside and outside the finger, vein on the back of the finger, etc.
- iris auricle
- facial vein subconjunctival blood vessel
- palm vein back vein
- palm print joint print inside and outside the finger
- vein on the back of the finger etc.
- An object of the present invention is to provide a biometric authentication system and a biometric authentication method that can achieve high-precision and high-speed authentication even when posture variation or shielding occurs during multimodal biometric imaging.
- a storage device for storing a plurality of biometric feature values associated with each user, a photographing device for photographing the biometric body, an image photographed by the photographing device is inputted
- the imaging device captures a first biometric image of a first user during a first period, which is different from the first period.
- the first user's second and third living bodies are imaged.
- the authentication processing device calculates a first feature amount from the first living body photographed during the first period, and calculates a second feature amount from the second living body and the third living body photographed during the second period.
- a feature amount and a third feature amount are respectively calculated, and the biometric feature amount for each user stored in the storage device, the first feature amount, the second feature amount, and the third feature amount are calculated.
- the user is authenticated by matching the
- FIG. 1 is a diagram showing the overall configuration of a biometric authentication system according to Example 1;
- FIG. 4 is a diagram illustrating an example of a functional configuration of a program stored in memory according to the first embodiment;
- FIG. 1 is a schematic diagram showing the configuration of a multimodal biometrics authentication system using a general-purpose front camera according to the first embodiment;
- FIG. 4 is a diagram illustrating an example of a processing flow of a registration processing unit of the biometric authentication system according to the first embodiment;
- FIG. 4 is a diagram illustrating an example of a processing flow of an authentication processing unit of the biometric authentication system according to the first embodiment;
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by guiding the face and fingers at the same time.
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by simultaneously holding a face and a finger, according to the first embodiment;
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by simultaneously holding a face and a finger, according to the first embodiment;
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by simultaneously holding a face and a finger, according to the first embodiment;
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by simultaneously holding a face and a finger, according to the first embodiment;
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multi
- FIG. 10 is an example of a screen transition diagram at the time of authentication of multimodal biometric authentication technology in which authentication is performed by simultaneously holding a face and a finger, according to the first embodiment;
- FIG. 4 is an explanatory diagram of one method of buffering and selection processing of facial feature amounts according to the first embodiment;
- FIG. 10 is an explanatory diagram of one method of buffering and selection processing of feature amounts of a face and fingers according to the first embodiment;
- FIG. 10 is an explanatory diagram of one method of feature pair generation based on feature amounts of a face and fingers according to the first embodiment;
- FIG. 4 is an explanatory diagram showing an example of a multimodal biometrics authentication technology based on alternate authentication using a face and fingers according to the first embodiment;
- FIG. 4 is an explanatory diagram showing an example of a multimodal biometrics authentication technology based on alternate authentication using a face and fingers according to the first embodiment
- FIG. 10 is a diagram showing an example of a processing flow of an authentication processing unit of a biometric authentication system capable of singly matching a face and fingers, according to the second embodiment
- FIG. 10 is an explanatory diagram showing an example of a multimodal biometric authentication technique omitting processing of a face or fingers according to a second embodiment
- a processor may be the subject of the processing to perform the processing while appropriately using storage resources (eg, memory) and/or interface devices (eg, communication ports).
- a main body of processing executed by executing a program may be a controller having a processor, a device, a system, a computer, or a node.
- the main body of processing performed by executing the program may be an arithmetic unit, and may include a dedicated circuit (for example, FPGA (Field-Programmable Gate Array) or ASIC (Application Specific Integrated Circuit)) that performs specific processing. .
- biometric features include finger veins, fingerprints, joint patterns, skin patterns, finger contour shapes, fatty lobule patterns, length ratios of each finger, finger width, finger area, melanin pattern, and palm veins. , palmprints, veins on the back of the hand, facial veins, ear veins, or anatomically different biological features such as the face, ears, and irises.
- FIG. 1A is a diagram showing an example of the overall configuration of a biometric authentication system 1000 using biometric features in this embodiment.
- the configuration of this embodiment may be configured not as an authentication system but as an authentication device in which all or part of the configuration is mounted in a housing.
- the authentication device may be a personal authentication device including authentication processing, or a finger image acquisition device or a finger feature image extraction device that performs authentication processing outside the device and specializes in acquiring a finger image.
- the embodiment may be a terminal.
- the authentication device includes an imaging unit that captures a living body, and an authentication processing unit that processes the captured image and performs biometrics authentication. It is a device that performs authentication processing, and a system including the authentication device is called a biometric authentication system or a biometric authentication system.
- a biometric authentication system includes a biometric device and a biometric authentication system.
- a biometric authentication system 1000 of this embodiment shown in FIG. 1A includes an input device 2 that is an imaging unit, an authentication processing device 10, a storage device 14, a display unit 15, an input unit 16, a speaker 17, and an image input unit 18.
- the input device 2 includes an imaging device 9 installed inside a housing, and may include a light source 3 installed in the housing.
- the authentication processing device 10 has an image processing function.
- the light source 3 is, for example, a light-emitting element such as an LED (Light Emitting Diode), and irradiates the face 4 and fingers 1 with light as the living body of the user in a certain area presented on the input device 2 .
- the light source 3 may emit various wavelengths, may emit light transmitted through the living body, or may emit light reflected by the living body.
- the imaging device 9 captures images of the finger 1 and face 4 presented to the input device 2 . At the same time, a living body such as the iris, the back of the hand, or the palm may be photographed.
- the imaging device 9 is an optical sensor capable of capturing light of a single wavelength or multiple wavelengths, may be a monochrome camera or a color camera, and can simultaneously capture ultraviolet light or infrared light in addition to visible light. It may be a multispectral camera. Also, a distance camera capable of measuring the distance to an object may be used, or a stereo camera configuration in which a plurality of the same cameras are combined may be used.
- the input device 2 may include multiple imaging devices. Furthermore, the finger 1 may be plural, and plural fingers of both hands may be included at the same time.
- the image input unit 18 acquires an image captured by the imaging device 9 in the input device 2 and outputs the acquired image to the authentication processing device 10 .
- various reader devices for example, a video capture board
- reading images can be used.
- the authentication processing device 10 is composed of a computer including a central processing unit (CPU) 11, memory 12, and various interfaces (IF) 13, for example.
- the CPU 11 executes programs stored in the memory 12 to implement functional units such as authentication processing.
- FIG. 1B is a diagram showing an example of the functional configuration of a program stored in the memory 12 for realizing each function of the authentication processing device 10.
- FIG. 1B is a diagram showing an example of the functional configuration of a program stored in the memory 12 for realizing each function of the authentication processing device 10.
- the authentication processing device 10 includes a registration processing unit 20 that registers in advance a biometric feature of an individual in association with a personal ID, and an authentication processing unit 20 that performs authentication based on the biometric feature extracted by imaging.
- An authentication processing unit 21 that outputs an authentication result as an authentication result
- a living body detection unit 22 that detects the position of a living body and removes unnecessary background from the input image, and a shooting control that shoots the presented living body under appropriate conditions.
- a quality judgment unit 24 for judging quality such as the image quality of a photographed living body and the posture of the living body, and a feature extraction for extracting biometric features by appropriately correcting the posture of the living body during registration processing and authentication processing.
- the memory 12 stores programs executed by the CPU 11 .
- the memory 12 temporarily stores images and the like input from the image input unit 18 .
- the interface 13 connects the authentication processing device 10 and an external device.
- the interface 13 is a device having ports and the like for connecting with the input device 2, the storage device 14, the display section 15, the input section 16, the speaker 17, the image input section 18, and the like.
- the interface 13 functions as a communication unit, and is for the authentication processing device 10 to communicate with an external device via a communication network (not shown).
- the communication unit is a device that performs communication according to the IEEE802.3 standard if the communication network is a wired LAN, and a device that performs communication according to the IEEE802.11 standard if the communication network 30 is a wireless LAN.
- the storage device 14 is composed of, for example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive), and stores user registration data and the like.
- the registration data is obtained during registration processing, is information for verifying users, and is stored in association with a plurality of biometric feature amounts for each user.
- the user identification information includes facial feature quantity, finger feature quantity, image such as finger vein pattern, and biological feature data associated with the registrant ID.
- the image of the finger vein pattern is an image of the finger vein, which is the blood vessel distributed under the skin of the finger, taken as a dark shadow pattern or a slightly bluish pattern.
- the feature data of the finger vein pattern is generated from the data obtained by converting the image of the vein part into a binary or 8-bit image, or from the coordinates of the feature points such as the bends, branches, and end points of the veins, or the luminance information around the feature points. It is data consisting of features, or data obtained by encrypting them and converting them into a state that cannot be deciphered.
- the display unit 15 is, for example, a liquid crystal display, and is an output device that displays the information received from the authentication processing device 10, the posture guidance information of the living body, and the posture determination result.
- the input unit 16 is, for example, a keyboard or a touch panel, and transmits information input by the user to the authentication processing device 10.
- the display unit 15 may have an input function such as a touch panel.
- the speaker 17 is an output device that transmits the information received from the authentication processing device 10 as an acoustic signal such as voice.
- FIG. 2 is a schematic diagram showing the configuration of a multimodal biometrics authentication system using a general-purpose front camera described in this embodiment.
- multimodal biometric authentication is performed using the biometric features of the face and fingers of the left hand when the user logs into the notebook PC.
- the user activates the authentication function to log in to the PC when performing work on the notebook PC.
- the user is often positioned in front of the notebook PC 41 , and in a general notebook PC 41 , the camera 9 is installed above the display 42 in order to easily photograph the user's face 4 .
- the camera 9 is installed so that the vicinity of the front surface of the display 42 can be photographed, that is, the user's face 4 and left hand 45 can be photographed as a whole.
- the user's face 4 is photographed near the center of the angle of view of the image of the camera 9 .
- the authentication system activates the camera 9 to photograph the biometric features of the user.
- a preview image 47 which is an image in which the face guide 43 and the finger guide 44 are overlaid on the display 42 .
- the guide display may be omitted. By displaying the guide as necessary, the user's authentication operation becomes easier and the effect of improving convenience can be obtained.
- the user aligns the face 4, left hand 45 and finger 1 to the displayed guide position while viewing the preview image 47.
- a guide message 46 may be displayed on the preview screen to clearly indicate to the user that each living body will be presented.
- the authentication system performs authentication based on a plurality of biometric features when detecting the presentation of the biometric body, and shifts the notebook PC 41 to a login state when the user can be determined to be a pre-registered user. Specific methods of registration and authentication are described in detail below.
- FIGS. 3 and 4 are diagrams showing an example of a schematic flow of registration processing and authentication processing of multimodal biometric authentication technology using a plurality of biometric features, respectively, described in this embodiment.
- This registration processing and authentication processing are realized by, for example, a program executed by the CPU 11 of the authentication processing device 10 described above.
- the present embodiment is described on the premise that the user's face and four fingers of the left hand are photographed, but the face may be a partial feature of the face rather than the entire face.
- the fingers it may be one finger or any other number of fingers, or may be fingers of a plurality of hands. It may also be any number of biometric features of any type, such as generally and widely known iris, veins, fingerprints, palm prints, etc., other than the face and fingers.
- the registration processing unit 20 is activated by the user's instruction for registration processing, and first, the authentication system displays a preview image on the display unit 15 to explain to the user that the biometric is to be registered (S301).
- the user can understand the flow of the registration process from the displayed preview image.
- the preview image allows the user to show the ideal way of holding the face and four fingers over, the procedure for holding the object over, or a sentence such as "first photograph the face, then hold the four fingers of the left hand".
- the flow of registration processing will be described. As a result, errors in registration operations can be reduced.
- the current camera image is displayed on the display unit 15 as a preview image so that the user can visually recognize that he or she is being photographed by the camera.
- the display may be performed on the entire screen, or may be displayed in a small size on a part of the screen.
- the camera image is horizontally reversed and displayed so that the left and right of the user and the left and right of the image match, thereby making it easier for the user to hold up his or her body.
- face detection processing which is preprocessing for registering face feature amounts, is performed (S302).
- a deep neural network is prepared in which the relationships between facial images, facial positions, and facial organs (facial parts or landmarks) are learned in advance.
- the landmarks of the face include, for example, the center of the eye, the tips of the inner and outer corners of the eye, the edge of the eyelid (eyeline), the tip of the nose, the left and right corners of the mouth, the center position between the eyebrows, and the like.
- a rectangle enclosing them can be defined as a face area.
- biometric features relating to the face are registered without recognizing that the user is photographing the face as the biometric feature.
- the user When logging into the PC as shown in FIG. 2, the user is usually positioned in front of the PC, and the biometric features of the face can be registered without the user being aware of it.
- the frame of the bounding box of the face may be displayed superimposed on the preview image so that the user can see that the face is being detected.
- a guide such as "please turn your face to the front” may be displayed.
- a face guide simulating the outline of the face may be displayed so that the user can visually understand that the face is currently being photographed and where the face is to be held.
- the photographing control unit 23 is activated, and face photographing control is performed to photograph the face while appropriately adjusting camera parameters such as camera exposure time, gain, white balance, and focus (S303).
- camera parameters such as camera exposure time, gain, white balance, and focus
- the exposure time of the camera is adjusted so as not to cause blown-out highlights or blocked-up shadows inside the detected face ROI image
- the focus of the camera is adjusted so that the focus of the camera matches the face.
- the white balance of the entire image is automatically adjusted based on methods such as the gray hypothesis, which assumes that the average color of the entire image is the color of the ambient lighting.
- the exposure time may be adjusted by adjusting the exposure time of the camera, but the exposure time may be adjusted by software such as weighted integration of pixel values of a plurality of consecutive image frames.
- the soft exposure adjustment method can partially correct the exposure of an image, and therefore has the advantage of being able to optimally correct the exposure independently for each living body, such as the face and a plurality of fingers.
- the acquired face ROI image is normalized (S304).
- normalization scaling is performed so that the size of the face (area of the face ROI, etc.) is constant. Generate, multiply by a constant so that the brightness of the face is a constant value, and so on.
- This normalization is a pre-processing performed to stabilize the result of facial feature extraction performed later.
- facial feature extraction is performed (S305).
- a deep neural network is used that inputs a facial ROI image and outputs an arbitrary fixed-length feature vector.
- the L1 distances of feature vectors obtained from multiple face ROI images of the same face are minimized with respect to each other, and the L1 distances of feature vectors obtained from face ROI images of different faces are maximized. learn to be
- a deep neural network is used to obtain the feature vector from the face ROI image.
- the feature amounts obtained from the same face image have a small L1 distance, and the feature amounts obtained from different face images are transformed with a large L1 distance. Therefore, whether or not the face images are the same can be evaluated based on the distance (dissimilarity) between the patterns.
- L1 distance is described here, any distance space may be used, and distance learning of feature amounts is generally performed in order to classify the person himself/herself and others. Examples of widely known specific methods include a method using Triplet Loss, and a method such as ArcFace that can realize distance learning simply by learning a general class classification problem.
- the quality value of the face image is calculated (S306).
- the size of the face is the size of the above-described face ROI image, and if it is small, it can be determined that the face is not captured in a sufficiently large size, so the quality value is reduced.
- the brightness of the face is obtained from the average brightness of the face ROI image before normalization, and it can be judged that the quality value is low if it is darker than expected, if it is too bright, or if there are many overexposed pixels.
- the three-dimensional rotation angle is estimated with reference to the landmark position generated in the average front face, and the pitching rotation angle, rolling rotation angle, and yaw rotation angle of the face are estimated.
- the quality is judged to be high, and when the value is large, it is judged that the face is not a frontal face and the quality value is low.
- the facial feature amount is extracted from the facial ROI image, and the similarity of the facial feature amount to temporally preceding and succeeding frame images is determined by round-robin. In this case, it is judged that the quality value of the face image is low because the extracted feature amount is unstable.
- Each of these evaluation items is quantified, weighted, summed up, and the values fused to obtain the final quality value of the face image.
- a deep neural network is used that inputs multiple face ROI images arranged in time series and outputs an arbitrary scalar value.
- a low value for example, 10
- a high value for example, 1000
- a high value can be obtained when a time-series image of a face ROI with a high tendency to succeed in authentication is input. Since the value obtained in this way becomes a higher value as it is suitable for authentication, it can be used as a quality value.
- This method does not require manual enumeration of each evaluation item as described above, and therefore has the advantage of improving development efficiency and easily increasing the correlation between the quality value and authentication success.
- an item for detecting blinks may be included in the face quality value evaluation items. Before the facial image is taken, guidance such as "Please blink a few times" is displayed, and a certain amount of time is set for the person to blink. If not, the face is considered to be a fake face such as a printed matter, and the quality value is lowered. As a result, it is possible to reject at least a fake face image in which blinking cannot be performed as having a low quality value.
- face registration determination is performed (S307). Since the quality value of the face image is the criterion for judging whether the image quality and facial posture are suitable for registration and authentication, the quality value can be used to judge whether the currently acquired facial feature quantity registration is suitable or not. Therefore, it is possible to set a predetermined threshold for the quality value of the face image, and determine that registration is possible when the threshold is exceeded.
- the quality value of the face image increases by accident, authentication may become unstable if the data is registered. It may be determined that the feature amount of the face can be registered when it exceeds a certain value by integrally.
- a finger guide is displayed on the screen so as to be overlaid on the preview image, which is the camera image of the user (S309). While checking the finger guide displayed on the screen and a preview image of himself/herself, the user raises his or her left hand to match the finger guide. At this time, a guide message such as "please hold your left hand over" may be displayed.
- finger detection processing is performed (S310).
- the finger is first separated from the background, and then an ROI image of the finger is obtained by cutting out the finger on a finger-by-finger basis.
- background separation processing a deep neural network that receives a camera image of a finger held as input and outputs a finger mask image in which only the finger area is 1 and other areas are 0, is applied to any input image.
- learning is performed so that a mask image of the finger is output, and the mask image of the finger is acquired by the network and the background is masked (removed).
- deep learning can be used to extract finger landmarks such as fingertips, finger roots, and finger joints. Based on this, the finger can be cut out in a rectangular shape.
- finger shooting control is performed (S311).
- the processing is the same as the above-described face photographing processing, so the description is omitted.
- the finger image is normalized (S312).
- finger image normalization there is a method of correcting the finger thickness and three-dimensional inclination to a constant value based on perspective projection transformation, and a method based on the landmarks performed in the face normalization as described above. method can be corrected.
- a posture correction process for normalizing the thickness and orientation of all detected fingers two points, the fingertip point of each finger and the finger crotch on both sides, are included in the inside, and the central axis of the finger is the image. It is assumed that an image is generated that is rotated so as to be parallel to the horizontal axis and scaled so that the finger width of each finger is a constant value. As a result, the directions and thicknesses of the fingers on the ROI image of all the fingers are unified.
- the feature extraction unit 26 is activated to extract the features of the finger (S313).
- Finger feature extraction can be performed in the same manner as facial feature extraction described above. Note that finger veins, fingerprints, joint prints, epidermal prints, melanin prints, fat prints, and the like may be independently extracted as finger feature quantities, or they may be mixed.
- filtering processing such as general edge enhancement filters, Gabor filters, and matched filters is used to emphasize biological features such as line pattern features of epidermis and blood vessels and spot features of fat lobules.
- Biological features can be obtained by binarizing or ternarizing the result.
- it may be obtained by a method of extracting brightness gradient features from key points such as SIFT (Scale-Invariant Feature Transform) features.
- SIFT Scale-Invariant Feature Transform
- the quality value of the finger image is calculated (S314).
- a method of detecting the finger posture by extracting information on the fingertips, finger roots, and finger widths of multiple fingers from the finger video and determining whether the finger posture at that time is appropriate. There is In the finger posture determination, based on the results of finger posture detection, it is confirmed that the finger is in an appropriate position by confirming that it does not deviate significantly from the displayed finger guide, and that the finger is stationary for a certain period of time. It is an evaluation item that there is
- the finger posture information such as the position of the fingertip does not change over time. Since it is difficult to keep the finger completely stationary, it may be determined that the finger is stationary when the movement amount is within a certain range. Even so, if the finger is not still or if the finger looks too far away (if the finger is far from the camera and the hand looks small), a guide will be displayed to that effect, Although not shown, the process may return to the process (S309) of urging the finger to be presented again.
- data suitability determination may be performed to detect whether the pattern extracted in this process is appropriate and whether the photographed finger is not a foreign object or a forgery. If the result of this determination is inappropriate, the quality value is greatly reduced so that it is not selected as a candidate.
- the data adequacy determination process it is difficult to extract a highly continuous pattern such as a blood vessel pattern even though it is a line feature, or a strong edge that cannot be observed with a real finger is extracted from the original image. If it is observed, it can be dismissed as a failure to extract the pattern or as an input forgery.
- a method may be used in which pulsations in image brightness due to changes in finger blood flow are detected from a moving image, and if pulsations cannot be detected, they are discarded.
- finger registration determination is performed (S315). This determination is performed based on the quality value of the finger image as described above, and the method can be performed in the same manner as the above-described method based on the quality value of the face image.
- the feature data of three registration candidates are collated in a brute-force manner to calculate the similarity between each candidate, and the sum of the similarities of the two candidates with the other candidates is the highest.
- the biometric body presented by the user is imaged, the feature amount of the biometric body is extracted, and compared with each feature data of the registered data.
- a registrant ID is output, and an authentication failure notice is output when there is no registration data that can be identified as the person himself/herself.
- the authentication processing unit 21 is activated by the user's instruction for authentication processing, and a preview image indicating that authentication has started is displayed (S401). For example, it displays "Left shooting will start”. The user can understand the flow of authentication processing from the displayed preview image.
- Face and hands are photographed for authentication, but as described in Figure 2 above, users are often positioned in front of terminals such as notebook PCs, so users should be especially aware of photographing their faces. Therefore, the user can be guided to shoot only the left hand. As a result, the user can prepare in advance to hold the left hand over the device, and the biometric authentication can be performed smoothly. In addition, the preview image of the camera is displayed in the same manner as in the registration process.
- the face detection process (S402) and the face image quality value calculation (S406) are performed for a predetermined period of time. These processes are the same as the face detection process (S302) to the facial image quality value calculation (S306) of the registration process in FIG.
- This series of processes (S402 to S408) is a process related to shooting of a single face, and here, the first period is called a "single face shooting phase”.
- the single face shooting phase is set for a certain period of time, but this phase is started when a certain number of high-quality face feature values are collected in the buffer, or when a single face is successfully compared with registered data. You can leave.
- the advantage of using a fixed period of time is that if the loop is repeated until facial feature values with high quality values are obtained, it will take too much time to get through this phase if the shooting proceeds in an environment where it is difficult to increase the quality values. There may be delays in authentication.
- this phase is limited to a fixed time, there is an advantage that it is difficult to delay.
- a finger guide is displayed (S409) to prompt the user to hold his or her finger over, and then face and finger detection processing (S410) is performed to calculate the quality values of the face image and the finger image. (S414) is performed.
- the facial feature amount is buffered (stored) in the memory 12 (buffer) of the authentication processing device 10, and the stored facial feature amount is selected (S415).
- the facial feature amount having a certain quality value is buffered, and the facial feature amount to be used in the matching process to be performed later is selected from the buffer.
- Facial feature quantities in the "single-face shooting phase” and “multi-shooting phase” are values obtained from images of the face. For example, if the captured image is a moving image with 30 frames per second, 30 facial feature amounts can be calculated per second. The same applies to the finger feature amount in the “multi-imaging phase”.
- the facial feature amount and the finger feature amount are fused, and the matching unit 26 is activated to acquire the image by the processing shown in FIG.
- Sequential collation of authentication data of face feature amount and finger feature amount with one or more registered data (usually, it is assumed that a plurality of registrants are registered) registered in advance in the storage device 14 to acquire a collation score (S417).
- the finger feature amount and the face feature amount are separated internally, a matching score is calculated as the degree of difference from each registered feature amount, and finally the matching scores are weighted and summed up.
- the finger feature amount and the face feature amount may be collated with registered data without being separated.
- matching with registered data may be confirmed based on information obtained by converting biometric feature amounts into encrypted feature amounts by, for example, PBI (Public Biometric Infrastructure) technology.
- Match scores may be scalar or vector, and the values may be binary or multi-valued.
- authentication is determined based on the calculated matching score (S418).
- a matching score for a single face feature amount and a matching score for a finger feature amount are obtained, and each of the matching results (dissimilarity) is below a threshold for recognizing similarity to a registrant.
- This judgment method requires that both biometric features are similar to the registrant, and has the effect of reducing the false acceptance rate of falsely judging an unregistered person as a registrant.
- score level fusion determination may be performed by linearly combining the matching scores of each living body, and furthermore, the matching scores of each living body are treated as a two-dimensional matching score vector, a threshold boundary hyperplane is defined in a multidimensional space, and the matching If the score vector is included in the area where the person can be identified, it may be determined that the data is similar to the registered data.
- the vector-based method can flexibly set the authentication threshold, and if there is a correlation between the matching scores of each biometric, the boundary can be defined according to the correlation, so highly accurate authentication judgment can be realized. It is possible.
- the degree of similarity between each of the extracted epidermal features and vein features and registered data may be calculated.
- 1:N authentication that determines a unique registrant from a plurality of registrants has been described, but 1:1 verification is performed by designating a registrant ID in advance and verifying that the registrant is that registrant before authenticating. Needless to say, it may be configured for authentication.
- the authentication determination (S418), it may be a requirement that the registration matches continuously. In this case, even if the registered data matches the registered data in the first authentication determination, the authentication is not successful, and the authentication is successful when the matching is confirmed a predetermined number of times or consecutive times. As a result, it is possible to prevent a false acceptance error in which a stranger accidentally succeeds in authentication, and it is possible to achieve stable and highly accurate authentication.
- the face and finger feature amounts are merged as in S417, but a method may be used in which the face alone is collated and the finger feature amounts are collated alone. This method will be described later with reference to FIG. do.
- An advantage of fusing both feature amounts is that it is possible to increase the strength against attacks by inputting forged many face images and finger images. If the authentication system is designed so that facial features and finger features can be individually verified, each matching result can be confirmed individually. Authentication can also be successfully attacked. On the other hand, if the features are fused so that they cannot be matched individually, the attack of multimodal authentication can only be successful if both are successfully attacked at the same time. is generated, the difficulty of the attack can be increased. In order to suppress such fraud, it is effective to use a feature level fusion method in which both feature quantities are merged and then verified.
- Figs. 5A to 5E are examples of screen transition diagrams during authentication of multimodal biometric authentication technology that performs authentication by guiding the face and fingers at the same time.
- a typical authentication failure that occurs when the face and fingers are guided simultaneously without providing the face-only shooting phase and the face image is not buffered.
- FIG. 5A is a preview image 47 immediately after the user activates the authentication screen to log in to the notebook PC.
- the user is often positioned in front of the terminal, and the camera 9 shoots the image centering around the front of the display 42, so the user's face 4 is shot near the center of the shot image.
- a face guide 43 and a finger guide 44 for holding the face and hands at the same time are displayed.
- a finger guide 44 is displayed on the left side of the screen and a face guide 43 is displayed on the right side of the screen in order to guide the user to hold the face and the left hand at the same time.
- the user is guided by a message 46 or the like.
- the user first holds the face over the correct position while checking the face 4 and the face guide 43 in the preview image 47 as shown in FIG. 5C.
- the user's left hand 45 is held over the finger guide 44 as shown in FIG. 5D.
- the user holds the left hand in front of the face, and at this time, the screen may be difficult to see because the user's left hand blocks the field of view. Therefore, in order to secure the field of view, the person may tilt the head and lay the face 4 sideways as shown in FIG. 5D. In this case, the face may not be detected correctly, or the feature amount of the face may fluctuate because the face is photographed in a posture different from that at the time of registration, making it difficult to recognize the registered face.
- the screen may be peeped through between the fingers of the left hand in order to secure the field of vision. Pictures cannot be taken.
- the color of the finger is often similar to the color of the face, if the face overlaps directly behind the finger, the boundary between the finger and the face becomes ambiguous, making it difficult to detect the finger. Detection may not be performed accurately. In either case, it becomes impossible to capture a face image of at least the same quality as the registered face image.
- Figs. 6A to 6E are examples of screen transition diagrams of multimodal biometric authentication technology for the face and fingers, including the single face imaging phase, as shown in Fig. 4 above.
- Fig. 4 the single face imaging phase
- the authentication system activates the camera 9 to capture the biometric image.
- the image at that time is presented to the user as a preview image 47, as in FIG. 5 described above.
- a guide message 46 is displayed for a certain period of time as a preview image 47 for displaying that authentication will be performed.
- the certain period of time is, for example, one second, and can be set to any number of seconds.
- the user does not need to align his/her face or hand with the guide, but this is the single face shooting phase shown in FIG.
- the value is calculated behind the scenes.
- the user is positioned in front of the notebook PC in the single-face shooting phase, and the user does not particularly hold his/her hand over the notebook PC. Therefore, the face is not dared to be tilted, and it is not blocked by the hand. Therefore, a high-quality face image can be taken without making the user aware of any operation.
- a finger guide 44 for holding the left hand is displayed as shown in FIG. 6C. Since the purpose of the multi-shooting phase is to shoot the left hand with high quality, display of the face guide may be omitted in this embodiment.
- the user holds the left hand over the correct position while checking the image of the user and the guide for the left hand as shown in FIG. 6D.
- the neck may be tilted sideways.
- a relatively high-quality face image is captured and buffered in the single face capturing phase, even if the face posture varies in this phase, there is no significant effect.
- multimodal biometric authentication is performed by combining the face image in the buffer and the finger image captured in the multi-photographing phase, and authentication can be performed with high-quality feature values for both.
- the authentication process can also be performed using the face image captured in the multi-photographing phase.
- facial images captured in the "single-face shooting phase" and “multi-shooting phase” can be used as information for authentication, and compared with the registered authentication data (facial features).
- both living bodies can be optimally imaged.
- both living bodies can be optimally imaged.
- the difference in the degree of blur between the images is used to determine the PSF ( By estimating the Point Spread Function, blurring of the entire image may be corrected, and an all-focus image in which all subjects are in focus may be generated. As a result, a plurality of biological features can be clearly photographed.
- the single face phase transitions to the next multi-photographing phase after a certain period of time. You can move to the next phase when you can confirm Alternatively, when a predetermined number of face images exceeding a predetermined quality value are collected, the process may proceed to the next phase.
- the next phase will be performed, at least while suppressing the deterioration of the recognition accuracy of the face recognition. It also has the advantage of shortening the shooting time.
- the drawback of the face-only matching method is the possibility that the information used to create the forgery may be leaked. Therefore, by always providing a single face phase for a certain period of time to prevent the user from guessing whether or not face recognition has succeeded, it is possible, for example, to determine whether at least single face recognition succeeds when a large number of forged face images are presented. Forgery can be made more difficult by making it impossible to guess from the behavior of the authentication system. Any method can be adopted according to the security policy of the authentication system.
- the finger guide 44 is displayed in the multi-photographing phase, but as another embodiment, the face guide may also be displayed on the right side. In that case, the effect of moving the face to the right is obtained.
- only the finger guide 44 may be displayed for a certain period of time to repeat multimodal authentication of the face and fingers, and if the authentication is not successful after the certain period of time has passed, the face guide may be additionally displayed. With this method, only the hand guide is displayed at first, so the user can focus on aligning the hand position. By displaying the face guide, it is expected that the user will move the position of the face away from the position of the hand. can increase the likelihood of
- FIGS. 7A to 7C are explanatory diagrams of one method of buffering and selection processing of facial feature amounts in FIG. 4 proposed in this embodiment.
- FIG. 7A is a graph plotting face feature amounts and their quality values in chronological order in the single face shooting phase, and schematically represents the part corresponding to the processing of S405 to S408 in FIG. 4 described above.
- the horizontal axis is time and the vertical axis is the quality value, showing the transition of the quality value when Ft is the facial feature value at time t.
- a threshold value for judging high quality is set, and feature amounts exceeding this threshold value are buffered.
- the facial feature amounts F3, F4, F5, F8, and F9 exceed the high quality threshold, and the facial feature amounts F3, F4, F5, F8, and F9 are selected to are stored in the buffer 141 in chronological order.
- the quality values can be plotted in chronological order for both the face feature amount and the finger feature amount.
- the facial feature amounts F10, F14, F17, and F18 are of high quality, and it can be seen that they are stored in the facial feature amount buffer 142 in the multi-photographing phase.
- the facial feature amounts F10, F14, F17, and F18 are also stored in the facial feature amount buffer 141 in chronological order.
- Ht indicates the finger feature amount, and here, H12, H15, H17, H18, and H19 exceed the high quality value.
- the buffers are clearly divided according to each of the face-only phase and the multi-photographing phase, but it goes without saying that the buffers may be managed by the same buffer. Further, in this embodiment, no buffer is provided for the finger feature amount, but the finger may be buffered in the same manner as the face, and the finger feature amount to be used may be selected according to the selection method described later.
- the face and finger feature amount selection processing performed in S415 and S416 of FIG. 4 is performed.
- the feature values of the face and fingers are not always of high quality. be.
- the design is such that authentication is performed only when both are of high quality, there is no particular need to buffer the feature amount. and 18, only two opportunities to authenticate occur. Therefore, the authentication is likely to fail.
- the face feature amount and the finger feature amount are selected from the buffered feature amounts and combined, that is, a feature pair (fused feature amount) is generated. This increases the chances of authentication, performs as many authentication processes as possible at the earliest possible timing, and increases the success rate of authentication.
- the quality value of the finger feature quantity is higher than the threshold, it is always a candidate for selection.
- the face feature quantity to be paired first select the face feature quantity from the buffer of the single face shooting phase, and then select from the buffer of the multi-shooting phase at the next opportunity.
- the buffer is alternately selected. In each buffer, selection is made in chronological order from the past to the new. However, if the most recently selected feature amount is chronologically continuous, there is a high possibility that both feature amounts are similar, so the next stored feature amount is selected.
- H12 is selected as the finger feature amount
- F3 which is stored most recently in the buffer for the single face shooting phase, is selected as the face feature amount paired with H12.
- the matching process is performed by pairing the finger feature amount of H12 and the face feature amount of F3. That is, using feature pairs, matching processing is performed based on the facial feature amount and finger feature amount for each user stored in the storage device.
- H15 is selected as the finger feature quantity. , and select F10 stored most recently.
- the face feature quantity paired with H18 is selected from the multi-imaging phase buffer, and since F10 was selected first, F14 is selected next this time. And finally H19 and F8 are selected as a pair.
- the variation of the feature quantity can be increased more than using a face feature quantity with a small amount of change.
- matching is performed by combining face feature amounts and finger feature amounts that are temporally different in timing. are observed at the same time (existing in the same image at the same time). Therefore, for example, at the time when the finger feature H17 is obtained in FIG. 7C, the facial feature F17 obtained at the same time is utilized, and the facial feature F5 to be combined with the finger feature H17 is the same person as F17 (similarity is high). As a result, if the face image at the current time is replaced with the face image taken in the single face shooting phase, it will be considered fraudulent and authentication will not be successful, making it a safer authentication system. can do.
- the buffered face feature values are arranged in chronological order, but they may be arranged in descending order of the quality value of the face image and used in that order. Since face images are selected in registration so that the quality value is as high as possible, if the quality value of the face image is high, it is highly likely that it is similar to the registration data. Therefore, by fusing facial features in descending order of quality, the probability of successful authentication can be increased as early as possible. At this time, the acquired time is also buffered in the same way as the feature quantity and quality value. can be selected. In addition, it is also possible to check the feature amounts of faces in the buffer against each other so that feature amounts with high similarity are not selected (thinning).
- the quality value includes a parameter related to face orientation, which may be selected so that the face orientation varies to some extent. For example, after selecting an image with the pitch angle of the face slightly upward from the front, an image with the face facing downward is preferentially selected next, or an image with the yaw angle of the face slightly to the left and right from the front are alternately extracted. It is also conceivable to do As a result, it is possible to exhaustively use the feature amounts of different face orientations for matching, so that the effect of increasing the success rate of authentication at an early stage can be obtained.
- any of the selection methods it is possible to avoid selecting facial features that are similar in succession, so that as many variations of facial features as possible can be used for authentication at an early stage, and the authentication success rate is increased at an early stage. It has a boosting effect.
- the facial feature amounts in the buffer may be reused in order from the beginning. At this time, if there are features that have not been selected as described above, they may be preferentially used. You can reuse it.
- feature pairs of authentication data face features, finger features
- registered face features face features
- finger features finger features
- the acquired finger features are obtained by combining the facial features acquired in the single-face imaging phase and the facial features acquired in the multi-imaging phase.
- Figs. 8A and 8B are explanatory diagrams of an example of multimodal biometric authentication technology based on alternate authentication using the face and fingers.
- the user simultaneously holds up their face and hands, performs face detection and facial feature extraction, finger detection and finger feature extraction, and performs multimodal biometric authentication. At this time, it is conceivable that the performance may be degraded due to simultaneous processing of the face and fingers.
- finger feature extraction is performed by skipping one frame.
- the horizontal axis represents the passage of time t
- Ft and Ht represent the facial feature amount and the finger feature amount
- the feature extraction of the fingers is processed for a maximum of four fingers, and since the matching is performed by brute-force matching one finger at a time, the processing time is longer than that for face matching.
- the extraction process of the finger feature amount is simplified every other frame by utilizing the temporal locality that the feature amount close in time is unlikely to change greatly.
- simplification of the process that is, speeding up of the authentication process can be realized without lowering the authentication accuracy as much as possible.
- the previous finger feature amount is reused when the time is an even number as in FIG. 8A, but in the case of the face feature amount, the previous facial feature amount is reused when the time is an odd number. do. Since each is processed in turn, new feature pairs are always generated. As a result, it is possible to reduce the average processing time by half while always generating new feature pairs, and to improve the perceived speed without degrading the authentication accuracy.
- FIG. 9 is an example of the processing flow of multimodal biometric authentication that independently matches the face and fingers. While FIG. 4 described above is an example of fusing face and finger patterns, FIG. 9 shows a processing flow in which matching is performed on a single face or a single finger, and the result is fused at the score level. Since the registration process can be performed in the same manner as in FIG. 3, the description is omitted.
- the face matching result may be a match score indicating similarity to the registered data, or may be a result of match or mismatch with the registered data determined by thresholding the match score.
- reset processing of the result according to the expiration date of the face matching result is performed (S909).
- the face matching result obtained at a certain time is held for a predetermined period of time, after which the result is invalidated.
- the fixed period of time during which the results are held is called an expiration date, and the invalidation of the results is called a reset.
- multimodal biometric authentication which is a method of independently matching multiple biometric features
- the success or failure of matching can be obtained for each of the multiple modals. It may hang.
- setting an expiration date for each verification result and once successful verification is always considered successful within the validity term, the possibility of successful authentication in all modals increases. This has the effect of increasing the success rate and shortening the processing time up to authentication.
- a successful verification is permanently enabled, for example, if another person accidentally succeeds in authentication in one modal, and the result is permanently enabled, erroneous false acceptance will occur. It becomes easier.
- a loop of processing (S911 to S917) for confirming that the quality values of the face and fingers are sufficiently high from the display of the finger guide is entered, but this is also basically the same processing as in FIG.
- processing for modals that have been successfully matched within the validity period at the present time is omitted. That is, at this time, if the facial feature amount is already sufficiently similar to the registered data, that is, if it can be determined that the face alone has been successfully authenticated within the validity period, face matching is omitted. Similarly, if a result that can be judged to be sufficiently similar to the registered data exists within the expiration date in the matching of the feature amount of the finger in the loop, the matching of the finger can be omitted.
- each matching score is recorded in chronological order (S918). ).
- each result is reset according to the expiration date of the matching result of the face and fingers (S919).
- authentication determination is performed by score level fusion using the collation score group obtained so far (S920).
- the process of authentication success is executed (S919, S921), and the authentication process is terminated. If the authentication fails, it is determined whether or not timeout has occurred (S922). If not, the process repeats from the display of the finger guide. If timeout occurs, authentication failure processing is performed (S923), and the authentication process ends.
- authentication determination by score level fusion using a group of matching scores performed in process S920, first, among the matching scores obtained in the past for the face and fingers within the expiration date, the degree of difference is the smallest.
- a fusion score is obtained by taking out each of the minimum scores and multiplying them by predetermined weights to sum them up, and if this falls below a predetermined threshold, score level fusion is performed to determine successful authentication. can also be adopted.
- score level fusion is a more suitable process than AND determination and OR determination because it can achieve higher-precision authentication.
- the authentication may be successful only when the scores are continuously below the authentication threshold, thereby suppressing accidental authentication acceptance errors of others.
- FIG. 10 is an explanatory diagram of an example of multimodal biometric authentication technology based on alternate authentication using the face and fingers in multimodal biometric authentication in which the face and fingers are independently matched.
- the single face shooting phase is limited to the face only, the amount of calculation is originally small, so the processing of F5 and F6 may be performed. In that case, for example, if the similarity is high even in F6, the facial feature processing up to F9 in the multi-photographing phase in the latter stage can be omitted, so that the amount of calculation in the latter stage can be further reduced.
- the multi-photographing phase is entered, and processing is performed for the face and fingers respectively.
- the finger feature quantity H7 has low similarity to the registered data, and the finger feature quantity H8 is subsequently processed.
- the facial feature amount F8 is also processed at the same time.
- the finger feature amount H8 has high similarity with registration, while the facial feature amount F8 has low similarity.
- the facial feature quantity F8 has low similarity
- the facial feature quantities F9 to F11 are again extracted and collated.
- the processing speed is improved because the focus can be placed on processing only facial features during this period. Assume that none of F8 to F11 are similar to the registered data.
- the present invention is not limited to the above-described embodiments, and includes various modifications.
- the above embodiments have been described in detail for better understanding of the present invention, and are not necessarily limited to those having all the configurations described.
Abstract
Description
2 入力装置
3 光源
4 顔
9 カメラ
10 認証処理部
11 中央処理部
12 メモリ
13 インターフェイス
14 記憶装置
15 表示部
16 入力部
17 スピーカ
18 画像入力部
20 登録処理部
21 認証処理部
22 生体検出部
23 撮影制御部
24 品質判定部
25 特徴抽出部
26 照合部
27 認証判定部
41 ノートPC
42 ディスプレイ
43 顔ガイド
44 指ガイド
45 左手
46 ガイドメッセージ
47 プレビュー画像
141 顔単体撮影フェーズの顔特徴量のバッファ
142 マルチ撮影フェーズの顔特徴量のバッファ
1000 生体認証システム
Claims (13)
- 生体を撮影する撮影装置と、前記撮影装置に接続され、利用者毎に複数の生体の特徴量を対応付けて記憶する記憶装置と、前記撮影装置により撮影した画像を入力し、入力した画像により生体認証を行う認証処理装置を含む認証システムにおいて、
前記撮影装置は、
第1の期間で、第1の利用者の第1の生体を撮影し、
前記第1の期間とは異なる第2の期間で、第1の利用者の第2の生体及び第3の生体とを撮影し、
前記認証処理装置は、
前記第1の期間において撮影した第1の生体から第1の特徴量を算出し、
前記第2の期間において撮影した第2の生体及び第3の生体から、第2の特徴量と第3の特徴量をそれぞれ算出し、
前記記憶装置に記憶された利用者毎の生体の特徴量と、前記第1の特徴量、前記第2の特徴量及び前記第3の特徴量とを照合することで利用者の認証を行う
ことを特徴とする認証システム。 - 請求項1に記載の認証システムにおいて、
前記記憶装置は、利用者毎に利用者識別情報と、第1の生体と、第2の生体の特徴量を対応して記憶しており、
前記第1の期間及び前記第2の期間で、前記撮影装置により撮影される、第1の生体と第2の生体は、同一利用者の同一領域であり、
前記認証処理装置が算出する第2の特徴量は、同一の利用者であって、第1の生体と異なる領域の特徴量であり、
前記記憶装置は、前記第1の期間に算出された第1の特徴量と、前記第2の期間に算出された第2の特徴量とを、複数記憶する
ことを特徴とする認証システム。 - 請求項2に記載の認証システムにおいて、
前記認証処理装置は、
前記第2の期間に算出された第3の特徴量と、前記第1の期間に算出された第1の特徴量、或いは、前記第2の期間に算出された第2の特徴量とで、特徴ペアを生成し、
前記特徴ペアを用いて、前記記憶装置に記憶された利用者毎の第1の生体の特徴量及び第2の生体の特徴量に基づいて照合処理を行う
ことを特徴とする認証システム。 - 請求項3に記載の認証システムにおいて、
前記認証処理装置は、
前記第1の特徴量、前記第2の特徴量に対する品質値を算出し、前記品質値が所定の値を超える場合に、前記第1の特徴量、前記第2の特徴量および対応する品質値を、前記記憶装置に記憶する
ことを特徴とする認証システム。 - 請求項1に記載の認証システムにおいて、
前記認証処理装置は、
前記第1の期間に算出された前記第1の特徴量及び前記第2の期間に算出された第2の特徴量と前記第2の期間に算出された第3の特徴量とを融合した融合特徴量を利用する
ことを特徴とする認証システム。 - 請求項4に記載の認証システムにおいて、
前記認証処理装置は、
前記第2の期間に算出された第3の特徴量に対する品質値が所定の値を超える第3の特徴量を選択し、
前記第1の期間に算出された第1の特徴量と、前記第2の期間に算出された第2の特徴量に対する品質値が所定の値を超える特徴量の内、前記第1の期間に算出された第1の特徴量と前記第2の期間に算出された第2の特徴量とを交互に選択して、前記選択された第3の特徴量と前記特徴ペアを生成する
ことを特徴とする認証システム。 - 請求項6に記載の認証システムにおいて、
前記記憶装置に、前記第1の期間に算出された第1の特徴量と前記第2の期間に算出された第2の特徴量に対する品質値が所定の値を超える特徴量を、時系列順に格納し、
前記認証処理装置は、前記特徴ペアを生成する際、前記第1の期間に算出された第1の特徴量と前記第2の期間に算出された第2の特徴量を、交互に古い順に優先して選択する
ことを特徴とする認証システム。 - 請求項6に記載の認証システムにおいて、
前記記憶装置は、前記第1の特徴量と前記第2の特徴量を品質値の高い順に格納し、
前記認証処理装置は、前記品質値の高い順に優先して特徴量を選択し、前記第3の特徴量とで特徴ペアを生成する
ことを特徴とする認証システム。 - 請求項6に記載の認証システムにおいて、
前記認証処理装置は、
前記第2の期間で、第2の特徴量の抽出と第3の特徴量の抽出とを交互に実施する、
ことを特徴とする認証システム。 - 請求項1に記載の認証システムにおいて、
前記認証処理装置は、
第1の特徴量と第2の特徴量とを独立に照合し、各照合結果を融合する
ことを特徴とする認証システム。 - 請求項10に記載の認証システムにおいて、
前記認証処理装置は、
前記第1の特徴量と前記第2の特徴量と、前記第3の特徴量とを独立に照合し、前記照合結果が前記記憶装置に予め登録された特徴量と類似度が高い場合は一定期間だけ当該特徴量の照合を割愛する
ことを特徴とする認証システム。 - 生体を撮影する撮影装置と、前記撮影装置に接続され、利用者毎に複数の生体の特徴量を対応付けて記憶する記憶装置と、前記撮影装置により撮影した画像を入力し、入力した画像により生体認証を行う認証処理装置を含む認証システムの生体認証方法において、
前記撮影装置は、
第1の期間で、第1の利用者の第1の生体を撮影し、
前記第1の期間とは異なる第2の期間で、第1の利用者の第2の生体及び第3の生体とを撮影し、
前記認証処理装置は、
前記第1の期間において撮影した第1の生体から第1の特徴量を算出し、
前記第2の期間において撮影した第2の生体及び第3の生体から、第2の特徴量と第3の特徴量をそれぞれ算出し、
前記記憶装置に記憶された利用者毎の生体の特徴量と、前記第1の特徴量、前記第2の特徴量及び前記第3の特徴量とを照合することで利用者の認証を行う
ことを特徴とする生体の認証方法。 - 請求項12に記載の生体の認証方法において、
前記認証処理装置は、
前記第2の期間に算出された第3の特徴量と、前記第1の期間に算出された第1の特徴量、或いは、前記第2の期間に算出された第2の特徴量とで、特徴ペアを生成し、
前記特徴ペアを用いて、前記記憶装置に記憶された利用者毎の第1の生体の特徴量及び第2の生体の特徴量に基づいて照合処理を行う
ことを特徴とする生体の認証方法。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US18/278,057 US20240126853A1 (en) | 2021-05-18 | 2022-02-18 | Biometric authentication system and authentication method |
EP22804278.4A EP4343689A1 (en) | 2021-05-18 | 2022-02-18 | Body part authentication system and authentication method |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2021-084231 | 2021-05-18 | ||
JP2021084231A JP2022177762A (ja) | 2021-05-18 | 2021-05-18 | 生体の認証システムおよび認証方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022244357A1 true WO2022244357A1 (ja) | 2022-11-24 |
Family
ID=84140535
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2022/006798 WO2022244357A1 (ja) | 2021-05-18 | 2022-02-18 | 生体の認証システムおよび認証方法 |
Country Status (4)
Country | Link |
---|---|
US (1) | US20240126853A1 (ja) |
EP (1) | EP4343689A1 (ja) |
JP (1) | JP2022177762A (ja) |
WO (1) | WO2022244357A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP7369316B1 (ja) * | 2023-02-27 | 2023-10-25 | 株式会社安部日鋼工業 | コンクリートを認証コードとする、識別システムおよび識別方法 |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004062846A (ja) | 2002-07-31 | 2004-02-26 | Waimachikku Kk | 個人識別装置の入力装置 |
JP2009020735A (ja) | 2007-07-12 | 2009-01-29 | Mitsubishi Electric Corp | 個人認証装置 |
JP2014211838A (ja) * | 2013-04-22 | 2014-11-13 | 富士通株式会社 | 生体認証装置、生体認証システム、および生体認証方法 |
WO2020065954A1 (ja) * | 2018-09-28 | 2020-04-02 | 日本電気株式会社 | 認証装置、認証方法および記憶媒体 |
WO2020065851A1 (ja) * | 2018-09-27 | 2020-04-02 | 日本電気株式会社 | 虹彩認証装置、虹彩認証方法および記憶媒体 |
JP2020095063A (ja) * | 2017-03-23 | 2020-06-18 | 株式会社Seltech | 個人認証装置および個人認証プログラム |
WO2020208824A1 (ja) * | 2019-04-12 | 2020-10-15 | 日本電気株式会社 | 情報処理装置、情報処理方法及び記録媒体 |
-
2021
- 2021-05-18 JP JP2021084231A patent/JP2022177762A/ja active Pending
-
2022
- 2022-02-18 EP EP22804278.4A patent/EP4343689A1/en active Pending
- 2022-02-18 WO PCT/JP2022/006798 patent/WO2022244357A1/ja active Application Filing
- 2022-02-18 US US18/278,057 patent/US20240126853A1/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004062846A (ja) | 2002-07-31 | 2004-02-26 | Waimachikku Kk | 個人識別装置の入力装置 |
JP2009020735A (ja) | 2007-07-12 | 2009-01-29 | Mitsubishi Electric Corp | 個人認証装置 |
JP2014211838A (ja) * | 2013-04-22 | 2014-11-13 | 富士通株式会社 | 生体認証装置、生体認証システム、および生体認証方法 |
JP2020095063A (ja) * | 2017-03-23 | 2020-06-18 | 株式会社Seltech | 個人認証装置および個人認証プログラム |
WO2020065851A1 (ja) * | 2018-09-27 | 2020-04-02 | 日本電気株式会社 | 虹彩認証装置、虹彩認証方法および記憶媒体 |
WO2020065954A1 (ja) * | 2018-09-28 | 2020-04-02 | 日本電気株式会社 | 認証装置、認証方法および記憶媒体 |
WO2020208824A1 (ja) * | 2019-04-12 | 2020-10-15 | 日本電気株式会社 | 情報処理装置、情報処理方法及び記録媒体 |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP7369316B1 (ja) * | 2023-02-27 | 2023-10-25 | 株式会社安部日鋼工業 | コンクリートを認証コードとする、識別システムおよび識別方法 |
Also Published As
Publication number | Publication date |
---|---|
US20240126853A1 (en) | 2024-04-18 |
JP2022177762A (ja) | 2022-12-01 |
EP4343689A1 (en) | 2024-03-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11188734B2 (en) | Systems and methods for performing fingerprint based user authentication using imagery captured using mobile devices | |
KR102561723B1 (ko) | 모바일 디바이스를 사용하여 캡처된 화상을 사용하여 지문 기반 사용자 인증을 수행하기 위한 시스템 및 방법 | |
KR102538405B1 (ko) | 생체 인증 시스템, 생체 인증 방법 및 프로그램 | |
WO2020190397A1 (en) | Authentication verification using soft biometric traits | |
JP4706377B2 (ja) | 生体判別装置および認証装置ならびに生体判別方法 | |
JP2009015518A (ja) | 眼画像撮影装置及び認証装置 | |
WO2022244357A1 (ja) | 生体の認証システムおよび認証方法 | |
WO2021166289A1 (ja) | データ登録装置、生体認証装置、および記録媒体 | |
Juluri et al. | SecureSense: Enhancing Person Verification through Multimodal Biometrics for Robust Authentication | |
US20210174068A1 (en) | Live facial recognition system and method | |
Al-Omar et al. | A Review On Live Remote Face Recognition and Access Provision Schemes | |
Singh et al. | Biometric Methods of Face Recognition: A Mirror Review | |
KR20210050649A (ko) | 모바일 기기의 페이스 인증 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22804278 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18278057 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022804278 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2022804278 Country of ref document: EP Effective date: 20231218 |