Biometric authentication device, biometric authentication method, and biometric authentication program

Abstract

A prediction unit predicts a stop position at which a body part of an authentication target person moving toward an imaging device stops, based on a plurality of images obtained by the imaging device capturing the body part in a distance measurement mode. A control unit switches a capturing mode of the imaging device from the distance measurement mode to an authentication mode before the body part stops at the stop position. An authentication unit performs a living body authentication for the authentication target person using an image of the body part captured by the imaging device in the authentication mode.

Description

Technical Field

[0001] This invention relates to liveness detection technology. Background Technology

[0002] Liveness detection is a technology that uses liveness features such as fingerprints, faces, and veins to verify identity. Palm vein authentication, which uses palm veins for verification, involves capturing vein patterns from images of the palm taken with a vein sensor and then using these patterns for authentication.

[0003] Regarding palm vein authentication, low-cost vein authentication devices that authenticate veins in the palm and fingers and require minimal installation space are known (e.g., see Patent Document 1). Automatic control devices that use a three-dimensional image sensor to predict the final indication coordinates based on an initial movement are also known (e.g., see Patent Document 2 and Non-Patent Document 1).

[0004] It is also known that a beam generating optical system can suppress the size of the light spot to a small size, prevent the reduction of the light spot brightness, and enable the camera device to be made thinner without reducing the sensitivity and accuracy of the ranging function (for example, see Patent Document 3).

[0005] Existing technical documents

[0006] Patent documents

[0007] Patent Document 1: Japanese Patent Application Publication No. 2008-246011

[0008] Patent Document 2: Japanese Patent Application Publication No. 2013-109444

[0009] Patent Document 3: International Publication No. 2018 / 078793

[0010] Non-patent literature

[0011] Non-patent document 1: Kusano, Omura, "Discussion on Jerk Minimalist Action Prediction Techniques Using Jerk's Finger", IPSJ Interaction 2012, p.929-934, 2012 Summary of the Invention

[0012] The problem that the invention aims to solve

[0013] In palm vein authentication, to capture a suitable image of the palm veins for authentication, the palm of the subject must be positioned within a specified range. This specified range is defined by upper and lower limits of coordinates that increase as the palm moves further away from the vein sensor. This specified range is sometimes also referred to as the authenticable range.

[0014] To capture images of palm veins within the authenticating range, ranging mode and authentication mode are sometimes used as the capture modes for vein sensors.

[0015] In ranging mode, a ranging image is obtained by photographing the palm using a ranging illumination light. This image is then used to measure the distance between the palm and the vein sensor, and the palm is guided so that its position is within the authentication range. Ranging mode is sometimes also called guiding mode. Authentication mode is a mode that obtains an authentication image of the palm veins by photographing the palm using an authentication illumination light.

[0016] However, in ranging mode, when switching from shooting mode to authentication mode after confirming that the palm has entered the authentication range and stopped, a delay occurs due to the switching. Therefore, it is difficult to quickly acquire palm vein images for palm vein authentication at the precise time when the palm stops within the authentication range.

[0017] This problem occurs not only in palm vein authentication, but also in various liveness authentication methods that use images of body parts.

[0018] In one aspect, the object of the present invention is to use images of body parts moving toward a camera device for rapid liveness detection.

[0019] Methods for solving problems

[0020] According to one embodiment, the liveness detection device includes a prediction unit, a control unit, and an authentication unit.

[0021] The prediction unit predicts the stopping position of the body parts based on multiple images obtained by the camera device in ranging mode capturing body parts moving towards the camera device. Before the body parts stop at the stopping position, the control unit switches the camera device's shooting mode from ranging mode to authentication mode. The authentication unit uses the images of the body parts captured by the camera device in authentication mode to perform liveness authentication on the subject.

[0022] According to another embodiment, in the liveness authentication method, the computer performs the following processing.

[0023] The computer predicts the stopping position of a body part as it moves toward the camera, based on multiple images captured by the camera device in ranging mode. Before the body part stops at the stopping position, the computer switches the camera device's shooting mode from ranging mode to authentication mode, and uses the images of the body part captured by the camera device in authentication mode to perform liveness authentication on the subject.

[0024] According to another implementation, the liveness authentication procedure causes the computer to perform the following processes.

[0025] The computer predicts the stopping position of a body part as it moves toward the camera, based on multiple images captured by the camera device in ranging mode. Before the body part stops at the stopping position, the computer switches the camera device's shooting mode from ranging mode to authentication mode, and uses the images of the body part captured by the camera device in authentication mode to perform liveness authentication on the subject.

[0026] Invention Effects

[0027] In one aspect, it is possible to quickly perform liveness detection using images of body parts moving toward a camera device. Attached Figure Description

[0028] Figure 1 This is a functional structure diagram of the liveness authentication device according to the implementation method.

[0029] Figure 2 This is the flowchart for the first liveness authentication process.

[0030] Figure 3 This is a functional structure diagram representing a specific example of a liveness detection device.

[0031] Figure 4 This is a diagram showing the illumination light used for ranging and the illumination light used for authentication.

[0032] Figure 5 It is a diagram showing the planar configuration of the light source for ranging and the light source for authentication.

[0033] Figure 6 It is a graph representing an image used for distance measurement.

[0034] Figure 7 This is a diagram representing the first switching control.

[0035] Figure 8 This is a diagram representing the second switching control.

[0036] Figure 9A This is the flowchart for the second liveness authentication process (Part 1).

[0037] Figure 9B This is the flowchart for the second liveness authentication process (Part Two).

[0038] Figure 9C This is the flowchart for the second liveness authentication process (section 3).

[0039] Figure 9D This is the flowchart for the second liveness authentication process (section 4).

[0040] Figure 10 This is a diagram representing the liveness authentication process.

[0041] Figure 11 This is a hardware structure diagram of an information processing device. Detailed Implementation

[0042] Hereinafter, the embodiments will be described in detail with reference to the accompanying drawings.

[0043] Figure 1 Example of the functional structure of a liveness detection device in an implementation method. Figure 1 The liveness detection device 101 includes a prediction unit 111, a control unit 112, and a detection unit 113.

[0044] Figure 2 It means Figure 1 A flowchart illustrating an example of the first liveness authentication process performed by the liveness authentication device 101. First, the prediction unit 111 predicts the stopping position of the body parts based on multiple images obtained by the camera device capturing body parts of the person being authenticated moving toward the camera device in ranging mode (step 201).

[0045] Next, before the body part stops at the stop position, the control unit 112 switches the shooting mode of the camera device from the ranging mode to the authentication mode (step 202). Then, the authentication unit 113 uses the image of the body part captured by the camera device in the authentication mode to perform liveness authentication on the person to be authenticated (step 203).

[0046] according to Figure 1 The liveness detection device 101 can quickly perform liveness detection using images of body parts moving toward the camera device.

[0047] Figure 3 express Figure 1 A specific example of the liveness detection device 101. Figure 3 The liveness detection device 301 includes a vein sensor 311, a prediction unit 312, a control unit 313, an authentication unit 314, and a storage unit 315. Figure 3 The prediction unit 312, control unit 313, and authentication unit 314 respectively correspond to Figure 1 The prediction department 111, the control department 112, and the authentication department 113.

[0048] The liveness authentication device 301 is used for various user authentication purposes in automated transaction devices such as ATMs (Automatic Teller Machines), access control devices, PCs (Personal Computers), multifunction printers, safes, lockers, etc.

[0049] The vein sensor 311 is a hardware camera device that captures an image of the palm veins of the person being authenticated. For example, the camera device described in Patent Document 3 can be used as the vein sensor 311.

[0050] The vein sensor 311 includes a near-infrared camera that illuminates the palm of the subject's hand with near-infrared light to obtain a near-infrared image of the palm as a palm vein image. The palm is an example of a body part of the subject.

[0051] The vein sensor 311 operates in either the ranging mode or the authentication mode, in the shooting mode. In the ranging mode, the vein sensor 311 illuminates the palm with a ranging illumination light called a ranging beam, and in the authentication mode, it illuminates the palm with an authentication illumination light.

[0052] Figure 4 Examples of ranging illumination light and authentication illumination light are shown. The user, acting as the authentication subject, raises their palm 401 above the vein sensor 311, moving the palm 401 towards the vein sensor 311. At this time, the vein sensor 311 operates in ranging mode, illuminating the palm 401 with ranging illumination light 402 and capturing an image of the palm 401 for ranging. The ranging illumination light 402 is, for example, visible light.

[0053] When the palm 401 enters the authentication range, the vein sensor 311 operates in authentication mode, illuminating the palm 401 with authentication illumination light 403 and capturing an image of the palm veins for authentication. The authentication illumination light 403 is, for example, near-infrared light.

[0054] A portion of the near-infrared light is reflected on the surface of the palm 401, but a portion penetrates through the skin and cells, seeping into the interior of the palm 401 and being absorbed by the hemoglobin contained in the blood of the veins. By photographing the palm 401 irradiated with near-infrared light using a near-infrared camera, it is possible to obtain an image of the palm veins, including not only the surface information of the palm 401 but also the vein pattern.

[0055] The vein sensor 311 continuously captures images for ranging or for authenticating palm veins at predetermined shooting intervals. Each captured image is sometimes referred to as a frame, and the shooting interval is sometimes referred to as the frame period.

[0056] Figure 5 This illustrates a planar configuration example of the light source for ranging and the light source for authentication in the vein sensor 311. Figure 5 The quadrilateral represents the shape of the vein sensor 311 when viewed from above. The ranging light sources 501-1 to 501-4 are arranged at the four corners of the quadrilateral, and the authentication light sources 502-1 to 502-16 are arranged in a circle inside the ranging light sources 501-1 to 501-4.

[0057] The rangefinding light sources 501-1 to 501-4 and the authentication light sources 502-1 to 502-16 are, for example, LED (Light-Emitting Diode) light sources. In rangefinding mode, rangefinding light sources 501-1 to 501-4 illuminate the palm 401 with rangefinding illumination 402. In authentication mode, authentication light sources 502-1 to 502-16 illuminate the palm 401 with authentication illumination 403.

[0058] Figure 6 An example of a range-measuring image is shown, obtained by photographing a hand 401 illuminated by a range-measuring illumination light 402. Figure 6 (a) represents an example of an image where the palm 401 is positioned close to the vein sensor 311. The four light spots 601 represent the ranging illumination lights 402 emitted from ranging light sources 501-1 to 501-4, respectively.

[0059] Figure 6 (b) represents an example of an image where the palm 401 is located far from the vein sensor 311. The four light spots 602 represent the ranging illumination lights 402 emitted from ranging light sources 501-1 to 501-4, respectively.

[0060] The prediction unit 312 checks whether four light spots are captured in the ranging image. If four light spots are captured, it determines that the palm 401 has been raised onto the vein sensor 311.

[0061] from Figure 6 The distance from the center of image (a) to each light spot 601 is greater than that of the other light spots. Figure 6 The distance from the center of the image in (b) to each spot 602 is long. Thus, a prescribed correlation exists between the distance from the center of the ranging image to each spot and the distances to the palm 401 and the vein sensor 311.

[0062] Therefore, the prediction unit 312 uses this correlation to determine the distance between the palm 401 and the vein sensor 311 at the time the ranging image was captured. The time of image capture is, for example, the time at which the image capture began. Then, based on the measured distance, the prediction unit 312 calculates the coordinate x(t(c)) of the palm 401 at the time of image capture indicated by index c.

[0063] The coordinate x(t(c)) represents the relationship with Figure 5 The coordinates are perpendicular to the plane. As coordinates x(t(c)), for example, the height representing the distance from the vein sensor 311 to the palm 401 is used.

[0064] When a hand 401 is detected to be raised, the prediction unit 312 generates starting point information 321 and stores it in the storage unit 315. The starting point information 321 includes c, t(c), and x(t(c)) at the starting point. At the starting point, c is 0, t(c) is 0, and x(t(c)) is x(0). Subsequently, whenever an image is captured, c and t(c) are updated using the following formula.

[0065] c = c + 1 (1)

[0066] t(c) = c × Δt (2)

[0067] Δt represents the shooting interval. c represents the number of shots taken between shooting time t(1) and shooting time t(c). The prediction unit 312 calculates the velocity V(t(c)) and acceleration a(t(c)) of the palm 401 at shooting time t(c) using the following formula.

[0068] V(t(c))={x(t(c))-x(t(c-1))} / Δt (3)

[0069] a(t(c))={V(t(c))-V(t(c-1))} / Δt (4)

[0070] The prediction unit 312 predicts the stopping position and stopping time of the palm 401 moving toward the vein sensor 311 based on a(t(c)) calculated from multiple images used for ranging and a motion model of the human body. The motion model of the human body is, for example, the Jerk minimal model described in Non-Patent Document 1.

[0071] The Jerk minimum model in non-patent literature 1 is a motion model that utilizes the case where the time integral of the jump is minimized when a person's hand moves. According to the Jerk minimum model, the velocity and acceleration are 0 at the start and end points of the movement of a person's fingertip. If the coordinates of the start point are set as x0 and the coordinates of the end point are set as xf, then the coordinates x(τ) of the fingertip can be expressed by the following formula using the parameter τ related to time t.

[0072] x(τ) = x0 + (xf - x0)(6τ) 5 -15τ 4 +10τ 3 (5)

[0073] τ=t / T (6)

[0074] T represents the time it takes for the fingertip to move from the starting point to the ending point. During this time, the velocity reaches a maximum only once, and the acceleration reaches a minimum after reaching a maximum. Therefore, if we define τ1 as the value when the acceleration reaches a maximum and τ2 as the value when the acceleration reaches a minimum, we can determine τ1 and τ2 based on the condition that the polynomial of the jump obtained by differentiating the right side of equation (5) three times is 0. The determined τ1 and τ2 are expressed by the following equation.

[0075] τ1 = (3 - 3) 1／2 ) / 6≒0.21 (7)

[0076] τ² = (3 + 3) 1／2 ) / 6≒0.79 (8)

[0077] As shown in equation (7), the time from the fingertip to the point where the acceleration reaches its maximum is approximately 21% of T. In other words, if we know the moment when the acceleration reaches its maximum when the palm 401 is moving, we can calculate the time T from the starting point to the ending point.

[0078] The endpoint is equivalent to the stopping position of hand 401. Therefore, the stopping time tf of hand 401 can be calculated based on the movement time T. First, in order to find the coordinate xf of the stopping position of hand 401, equation (5) is transformed to obtain the following equation.

[0079] xf=(x(τ)-x0) / (6τ 5 -15τ 4 +10τ 3 )+x0 (9)

[0080] Next, substitute x(τ1) into x(τ) in equation (9), and substitute τ1 into τ in equation (7) to obtain the following equation.

[0081] xf=(4x(τ1)-(2+3 1 / 2 )x0) / (2-3 1 / 2 (10)

[0082] When the acceleration a(t(c)) reaches its maximum, the acceleration a(t(c)) changes from increasing to decreasing. Therefore, the prediction unit 312 uses x(0) contained in the starting point information 321 as x0, and uses the coordinate x(t(c)) when the acceleration a(t(c)) changes from increasing to decreasing as x(τ1), and calculates xf using equation (10), thereby predicting the stopping position. When the slope of the acceleration a(t(c)) changes from positive to negative, the prediction unit 312 determines that the acceleration a(t(c)) has changed from increasing to decreasing.

[0083] Next, the prediction unit 312 checks whether xf is included in the verifiable range. If xf is below the upper limit and above the lower limit of the verifiable range, the prediction unit 312 determines that xf is included in the verifiable range.

[0084] When xf is within the verifiable range, the prediction unit 312 calculates the moment t1 when the acceleration a(t(c)) changes from increasing to decreasing using the following formula.

[0085] t1=c1×Δt (11)

[0086] c1 represents the index c when the acceleration a(t(c)) changes from increasing to decreasing.

[0087] The starting point information 321 contains t(c) which is 0, therefore the starting time is 0. Therefore, the time from when the palm 401 starts moving until the acceleration becomes maximum is the same as time t1, and the movement time T of the palm 401 is the same as the stopping time tf. Therefore, the prediction unit 312 uses equations (6) and (7) to calculate the stopping time tf using the following formula, thereby predicting the stopping time tf.

[0088] tf=t1 / 0.21 (12)

[0089] Then, the prediction unit 312 outputs the predicted stopping time tf to the control unit 313. The control unit 313 uses the stopping time tf output from the prediction unit 312 to calculate the number of shots N taken between time t1+Δt and the stopping time tf using the following formula.

[0090] N=(tf-t1) / Δt (13)

[0091] When the right side of equation (13) is not an integer, the prediction unit 312 calculates the integer N by discarding the value below the decimal point.

[0092] Next, the control unit 313 decrements N by 1 each time an image is captured, until N becomes a predetermined number M. M is an integer greater than or equal to 1 and less than N. Then, when N = M, the control unit 313 switches the imaging mode of the vein sensor 311 from the ranging mode to the authentication mode, thereby controlling the vein sensor 311 to capture images of the palm 401 in the authentication mode.

[0093] The moment when N = M corresponds to a predetermined time before the stop time tf. Therefore, by checking whether N becomes M, the control unit 313 can determine the predetermined time before the stop time tf and can switch the imaging mode of the vein sensor 311 from the ranging mode to the authentication mode at the predetermined time.

[0094] After switching the shooting mode from ranging mode to authentication mode, the vein sensor 311 acquires a palm vein image 322 for authentication by photographing the palm 401. The storage unit 315 stores the acquired palm vein image 322.

[0095] The authentication unit 314 uses the palm vein image 322 to perform palm vein authentication on the applicant. The storage unit 315 pre-stores a registration template 323. The registration template 323 includes registration feature information of one or more registrants. The authentication unit 314 obtains the registration feature information of each registrant, for example, from a database server not shown in the figure, and uses the obtained registration feature information to generate the registration template 323.

[0096] The authentication unit 314 extracts feature information of the vein pattern from the vein pattern contained in the palm vein image 322, and compares the extracted feature information with the registration feature information contained in the registration template 323.

[0097] For example, the authentication unit 314 calculates the similarity between the feature information of the vein pattern and the registration feature information of each registrant. If the similarity for any registrant is greater than a threshold, the authentication unit 314 determines that the authentication is successful; if the similarity for any registrant is less than the threshold, the authentication is deemed to have failed. Then, the authentication unit 314 outputs an authentication result indicating whether the authentication was successful or failed.

[0098] Figure 7 express Figure 3 An example of the first switching control in the liveness authentication process performed by the liveness authentication device 301. Figure 7 In the switching control, the shooting mode is not switched based on the stopping time tf predicted by the prediction unit 312, but rather after confirming that the palm 401 has entered the verifiable range 701 of the vein sensor 311 and stopped. The verifiable range 701 is defined by the upper limit 711 and the lower limit 712 of the coordinate x(t(c)).

[0099] The person being authenticated raises their palm 401 above the vein sensor 311, bringing the palm 401 close to the vein sensor 311. At this time, the vein sensor 311 takes a picture of the palm 401 in ranging mode, obtaining a ranging image represented by a white quadrilateral.

[0100] After confirming that the palm 401 has stopped within the authentication range 701, the control unit 313 switches the shooting mode from the ranging mode to the authentication mode. Thus, the vein sensor 311 captures an image of the palm 401 in authentication mode, obtaining an image of the authenticated palm veins represented by a diagonal quadrilateral.

[0101] However, due to the delay caused by the accompanying switch, the shooting mode has not yet switched to authentication mode when the palm 401 stops just below the upper limit value 711. Therefore, at the time when the palm 401 stops, the palm 401 is still being photographed in ranging mode, and palm vein authentication is not performed.

[0102] Figure 8 express Figure 3 An example of the second switching control in the liveness authentication process performed by the liveness authentication device 301. Figure 8 In the switching control, the shooting mode is switched based on the stop time tf predicted by the prediction unit 312.

[0103] and Figure 7 Similarly, the person being authenticated raises their palm 401 above the vein sensor 311, bringing their palm 401 close to the vein sensor 311. When the prediction unit 312 detects that the palm 401 has been raised, it calculates the coordinates x(0) of the starting point. The vein sensor 311 takes a picture of the palm 401 in ranging mode, obtaining a ranging image represented by a white rectangle.

[0104] When the prediction unit 312 detects that the acceleration a(t(c)) changes from increasing to decreasing, it uses the coordinate x(t1) at this time to calculate xf using equation (10) and calculates the stopping time tf.

[0105] The control unit 313 calculates the number of times N is captured between time t1+Δt and stop time tf using equation (13), and decrements N by 1 each time an image is captured. Then, when N=M, the control unit 313 switches the capturing mode from ranging mode to authentication mode. As a result, the vein sensor 311 captures the palm 401 in authentication mode, obtaining an authentication palm vein image represented by a quadrilateral with diagonal lines.

[0106] When M=1, even if the palm 401 stops just below the upper limit 711, the shooting mode can be switched to authentication mode to capture the palm 401 at the timing of the palm 401 stopping. This allows for rapid acquisition of palm vein images for palm vein authentication, thus improving the operability of the vein sensor 311 for the user being authenticated.

[0107] Furthermore, by setting the specified number M to an appropriate value, the palm 401 can be photographed by switching the shooting mode to authentication mode at a time when the palm 401 enters the authentication range 701. In this case, since palm vein authentication can be performed by quickly acquiring palm vein images, the operability of the vein sensor 311 by the person being authenticated is improved.

[0108] The authenticable range 701 and the shooting interval Δt vary according to the vein sensor 311. Therefore, by repeatedly conducting experiments on palm vein authentication using the liveness authentication device 301, the value of M can be determined, which is a value suitable for determining the timing when the palm 401 enters the authenticable range 701.

[0109] Additionally, when the palm 401 enters the authentication range 701 during a timed switching shooting mode, it is possible to acquire a palm vein image before the palm 401 stops, thus enabling palm vein authentication. However, even if the palm 401 is moving, as long as the speed of the palm 401 is sufficiently small, a suitable palm vein image for authentication can be acquired.

[0110] Figures 9A to 9D It means Figure 3 A flowchart illustrating an example of the second liveness authentication process performed by the liveness authentication device 301. In the following description, it will sometimes be referred to as... Figure 5 The range measuring light sources 501-1 to 501-4 are uniformly referred to as the range measuring light source group, and the certification light sources 502-1 to 502-16 are uniformly referred to as the certification light source group.

[0111] First, the control unit 313 controls the vein sensor 311 to turn off the ranging light source group and the authentication light source group (step 901), and controls the vein sensor 311 to operate in ranging mode (step 902). As a result, the vein sensor 311 illuminates the ranging light source group.

[0112] Next, the vein sensor 311 begins capturing an image (step 903) and obtains the captured image (step 904). Then, the prediction unit 312 uses the captured image to check whether the palm is raised onto the vein sensor 311 (step 905). If the palm is not raised (step 905, no), the liveness detection device 301 repeats the processing of step 903 and subsequent steps.

[0113] When the palm is raised (step 905, yes), the prediction unit 312 sets the index c to 0 (step 906). Next, the prediction unit 312 uses the captured image to determine the distance between the palm and the vein sensor 311, and calculates the coordinates x(t(c)) of the palm based on the measured distance (step 907).

[0114] Then, the prediction unit 312 generates starting information 321 containing c, t(c), and x(t(c)) (step 908). The starting information 321 contains c as 0, t(c) as 0, and x(t(c)) as x(0).

[0115] Next, the vein sensor 311 begins capturing the next image (step 909) and acquires the captured image (step 910). Then, the prediction unit 312 increments c by 1 (step 911), and the vein sensor 311 begins capturing the next image (step 912).

[0116] Next, the prediction unit 312 uses the image obtained in step 910 to calculate the coordinates x(t(c)), velocity V(t(c)), and acceleration a(t(c)) (step 913). Then, the prediction unit 312 checks whether x(t(c)) is within the verifiable range (step 914). If x(t(c)) is below the upper limit of the verifiable range but above the lower limit, the prediction unit 312 determines that x(t(c)) is within the verifiable range.

[0117] When x(t(c)) is within the authenticable range (step 914, yes), the vein sensor 311 acquires the captured image (step 915), and the prediction unit 312 requests the control unit 313 to switch the shooting mode of the vein sensor 311. Then, the control unit 313 switches the shooting mode from the ranging mode to the authentication mode (step 916). As a result, the vein sensor 311 turns off the ranging light source group and turns on the authentication light source group.

[0118] Next, the vein sensor 311 begins capturing images of the palm in authentication mode (step 917), acquiring a palm vein image 322 (step 918). Then, the authentication unit 314 uses the palm vein image 322 to perform palm vein authentication on the person being authenticated and checks whether the authentication was successful (step 919).

[0119] If authentication fails (step 919, No), the authentication unit 314 outputs an authentication result indicating authentication failure (step 921), and the liveness authentication device 301 repeats step 901 and subsequent steps. If authentication succeeds (step 919, Yes), the authentication unit 314 outputs an authentication result indicating successful authentication (step 920), and the liveness authentication device 301 ends the process.

[0120] When x(t(c)) is not included in the verifiable range (step 914, no), the prediction unit 312 checks whether the acceleration a(t(c)) changes from increasing to decreasing (step 922). If the acceleration a(t(c)) does not change from increasing to decreasing (step 922, no), the liveness authentication device 301 repeats the processing of step 910 and subsequent steps.

[0121] When the acceleration a(t(c)) changes from increasing to decreasing (step 922, yes), the prediction unit 312 performs the processing in step 923. In step 923, the prediction unit 312 uses x(0) contained in the starting point information 321 and the coordinate x(t(c)) when the acceleration a(t(c)) changes from increasing to decreasing to calculate the coordinate xf of the stopping position of the palm.

[0122] Next, the prediction unit 312 checks whether xf is included in the scope of authentication (step 924). If xf is not included in the scope of authentication (step 924, no), the liveness authentication device 301 repeats the processing of step 904 and subsequent steps.

[0123] When xf is within the verifiable range (step 924, yes), the prediction unit 312 calculates the moment t1 when the acceleration a(t(c)) changes from increasing to decreasing, and uses t1 to calculate the stopping time tf when the hand stops (step 925). Then, the prediction unit 312 outputs the predicted stopping time tf to the control unit 313.

[0124] Next, the control unit 313 uses time t1 and stop time tf to calculate the number of shots N taken between time t1+Δt and stop time tf (step 926).

[0125] Next, the vein sensor 311 acquires the image that has already been captured (step 927) and begins capturing the next image (step 928). Then, the control unit 313 decrements N by 1 (step 929) and compares N with a predetermined number M (step 930). If N and M are inconsistent (step 930, no), the liveness detection device 301 repeats step 927 and subsequent steps.

[0126] When N and M are the same (step 930, yes), the vein sensor 311 acquires the image that has been captured (step 931), and the control unit 313 switches the shooting mode from ranging mode to authentication mode (step 932). As a result, the vein sensor 311 turns off the ranging light source group and turns on the authentication light source group.

[0127] Next, the control unit 313 sets the number of retries R to 0 (step 933) and increments R by 1 (step 934). Then, the control unit 313 compares R with the threshold TH (step 935). For example, M+1 is used as TH.

[0128] If R is less than TH (step 935, Yes), the vein sensor 311 begins capturing the palm in authentication mode (step 936), acquiring the captured palm vein image 322 (step 937). Then, the authentication unit 314 uses the palm vein image 322 to perform palm vein authentication on the authentication subject and checks whether the authentication is successful (step 938).

[0129] If authentication fails (step 938, No), the authentication unit 314 outputs an authentication result indicating failure (step 940), and the liveness authentication device 301 repeats step 934 and subsequent steps. If R reaches TH (step 935, No), the liveness authentication device 301 repeats step 901 and subsequent steps. If authentication succeeds (step 938, Yes), the authentication unit 314 outputs an authentication result indicating success (step 939), and the liveness authentication device 301 terminates the process.

[0130] Figure 10 express Figure 3 An example of a liveness authentication process performed by the liveness authentication device 301. In this example, Δt = 10 milliseconds and M = 1.

[0131] Based on the image captured at 0 milliseconds, it was detected that the palm was raised. At c=3, based on the image captured at 20 milliseconds, it was detected that the acceleration a(t(c)) changed from increasing to decreasing. Therefore, c1=3, and according to equation (11), t1=30 milliseconds.

[0132] At this time, the prediction unit 312 uses the coordinate x(t1) to calculate xf using equation (10), and calculates the stopping time tf. In this example, tf = 143 milliseconds. The control unit 313 calculates the number of shots N taken between the next shooting time 40 milliseconds and the stopping time 143 milliseconds using equation (13). In this case, N = 11.

[0133] Next, the control unit 313 decrements N by 1 each time an image is captured. After the image capture begins at 130 milliseconds, N = M = 1. Therefore, the control unit 313 switches the capture mode of the vein sensor 311 from the ranging mode to the authentication mode.

[0134] Then, the vein sensor 311 begins capturing images of the palm 140 milliseconds before the stopping time 143 milliseconds in authentication mode, acquiring a palm vein image. The authentication unit 314 uses the acquired palm vein image to perform palm vein authentication. Thus, palm vein images can be acquired quickly at the precise time the palm stops for palm vein authentication.

[0135] On the other hand, when M=2, after the image capture begins at 120 milliseconds, N=M=2, so the shooting mode switches from ranging mode to authentication mode. Then, at 130 milliseconds, the palm capture begins in authentication mode. Thus, palm vein images can be rapidly acquired at precise intervals before the palm stops capturing for palm vein authentication. M can also be an integer greater than 3.

[0136] As another switching control, the liveness detection device 301 can also switch the shooting mode at a time when it is determined that xf is within the authenticable range. In this case, [the following is omitted]. Figure 9C Step 925~ Figure 9D The process proceeds to step 930. Then, if xf is within the authenticable range (step 924, yes), the prediction unit 312 requests the control unit 313 to switch the imaging mode of the vein sensor 311, and the liveness authentication device 301 immediately performs the processing of step 931 and subsequent steps.

[0137] In this case, palm vein images can be quickly acquired at a time before the hand stops moving, and palm vein authentication can be performed.

[0138] When using a timed switching shooting mode where the xf is determined to be within the authentication range, there is a possibility of obtaining a palm vein image and performing palm vein authentication before the palm enters the authentication range. However, if the palm movement is slow enough, a suitable palm vein image for authentication can be obtained.

[0139] Figure 1 The liveness detection device 101 and Figure 3 The structure of the liveness detection device 301 is merely an example; some components may be omitted or modified depending on the purpose or conditions of the liveness detection device. For example, in Figure 3 When a vein sensor 311 is externally mounted on the liveness detection device 301, the vein sensor 311 can be omitted from the structure of the liveness detection device 301. The vein sensor 311 can also capture images of body parts other than the palm to obtain vein images.

[0140] The liveness detection device 301 can replace palm vein detection, using other liveness images such as palm prints, fingerprints, and faces for liveness detection. In this case, an image sensor that acquires other liveness images is used instead of the vein sensor 311.

[0141] Figure 2 as well as Figures 9A to 9D The flowchart shown is just an example, and some parts of the process can be omitted or changed depending on the structure or conditions of the liveness detection device.

[0142] Figure 4 The ranging illumination and authentication illumination shown are merely examples, and the ranging illumination and authentication illumination vary depending on the vein sensor 311. Figure 5 The planar configuration of the distance measuring light source and the authentication light source shown is as follows: Figure 6 The distance measurement image shown is just one example; the planar configuration of the distance measurement light source and the authentication light source, as well as the distance measurement image, vary depending on the vein sensor 311.

[0143] Figure 7 , Figure 8 as well as Figure 10 The switching control shown is just an example; the switching control changes based on the predicted stopping position of the hand and the stopping time.

[0144] Equations (1) to (13) are merely examples; the liveness detection device 301 can also use other calculation formulas for switching control. The prediction unit 312 can also replace the Jerk minimum model and predict the stopping position and stopping time of the hand based on other motion models such as the torque change minimum model and the muscle tension change minimum model.

[0145] Figure 11 It shows the use of Figure 1 The liveness detection device 101 and Figure 3 Example of the hardware structure of the information processing device (computer) of the liveness authentication device 301. Figure 11 The information processing device includes a CPU 1101, a memory 1102, an input device 1103, an output device 1104, an auxiliary storage device 1105, a media drive device 1106, and a network connection device 1107. These components are hardware and are interconnected via a bus 1108. Figure 3 The vein sensor 311 can also be connected to the bus 1108.

[0146] The memory 1102 is, for example, a semiconductor memory such as ROM (Read Only Memory) or RAM (Random Access Memory), which stores programs and data used for processing. The memory 1102 can also be used as... Figure 3 The storage unit 315 operates.

[0147] CPU 1101 (processor) executes programs, for example, by utilizing memory 1102, as Figure 1 The prediction unit 111, control unit 112, and authentication unit 113 operate. The CPU 1101 executes programs using the memory 1102, and also acts as... Figure 3 The prediction unit 312, control unit 313, and authentication unit 314 operate.

[0148] Input device 1103, such as a keyboard or indicator device, is used for inputting instructions or information from the operator. Output device 1104, such as a display device, printer, or speaker, is used for outputting queries or instructions to the operator and for outputting processing results. The processing result may also be an authentication result indicating successful or unsuccessful authentication.

[0149] Auxiliary storage device 1105 may be, for example, a disk drive, optical disk drive, optical disc drive, magnetic tape drive, etc. Auxiliary storage device 1105 may also be a hard disk drive or an SSD (Solid State Drive). The information processing device can store programs and data in auxiliary storage device 1105 and load them into memory 1102 for use. Auxiliary storage device 1105 can also be used as... Figure 3 The storage unit 315 operates.

[0150] The media drive device 1106 drives the portable recording medium 1109 to access its recorded content. The portable recording medium 1109 can be a storage device, floppy disk, optical disk, magneto-optical disk, etc. The portable recording medium 1109 can also be a CD-ROM (Compact Disk Read Only Memory), DVD (Digital Versatile Disk), USB (Universal Serial Bus) storage device, etc. The operator can store programs and data in the portable recording medium 1109 and load the programs and data into the memory 1102 for use.

[0151] As described above, the computer-readable recording medium storing the programs and data for processing is a physical (non-transitory) recording medium such as memory 1102, auxiliary storage device 1105, or portable recording medium 1109.

[0152] The network connection device 1107 is a communication circuit that connects to communication networks such as WAN (Wide Area Network) and LAN (Local Area Network) to perform data conversion accompanying communication. The information processing device can receive programs and data from external devices via the network connection device 1107 and load them into the memory 1102 for use.

[0153] In addition, the information processing device does not need to include Figure 11All components may be omitted depending on the purpose or conditions of the information processing device. For example, if an interface with the operator is not required, the input device 1103 and the output device 1104 may be omitted. If a portable recording medium 1109 or a communication network is not used, the media drive device 1106 or the network connection device 1107 may also be omitted.

[0154] The disclosed embodiments and their advantages have been described in detail, but those skilled in the art can make various changes, additions, and omissions without departing from the scope of the invention as expressly stated in the claims.

Claims

1. A liveness detection device, characterized in that, The liveness authentication device includes: The prediction unit predicts the stopping position of the body parts based on multiple images obtained by the camera device capturing body parts of an authenticator moving toward the camera device in ranging mode. The control unit, before the body part stops at the stop position, switches the shooting mode of the camera device from the ranging mode to the authentication mode; and The authentication department uses the images of the body parts captured by the camera device in the authentication mode to perform liveness authentication on the person to be authenticated.

2. The liveness detection device according to claim 1, characterized in that, Based on the multiple images, the prediction unit calculates the position and acceleration of the body part at the time of each of the multiple images being captured, and predicts the stopping position based on the motion model of the human body and the position of the body part when the acceleration of the body part becomes maximum.

3. The liveness detection device according to claim 2, characterized in that, The prediction unit checks whether the stopping position is within the authenticable range. If the stopping position is within the authenticable range, based on the motion model, it predicts the stopping time of the body part at the stopping position, according to the moment when the acceleration of the body part becomes maximum. Based on the stop time, the control unit determines a predetermined time before the stop time, and at the predetermined time, switches the shooting mode of the camera device from the ranging mode to the authentication mode.

4. The liveness detection device according to claim 2, characterized in that, The prediction unit checks whether the stopping position is within the authenticable range. When the stop position is within the authenticable range, the control unit switches the shooting mode of the camera device from the ranging mode to the authentication mode.

5. The liveness detection device according to any one of claims 1 to 4, characterized in that, The ranging mode is a shooting mode that uses ranging illumination light to photograph the body part, and the authentication mode is a shooting mode that uses authentication illumination light to photograph the body part.

6. A liveness authentication method, characterized in that, The computer performs the following processing: Based on multiple images obtained by the camera device capturing body parts of an authenticator moving toward the camera device in ranging mode, the stopping position of the body parts is predicted. Before the body part stops at the stopping position, the shooting mode of the camera device is switched from the ranging mode to the authentication mode; as well as The image of the body part captured by the camera device in the authentication mode is used to perform liveness authentication on the person being authenticated.

7. A liveness authentication procedure for causing a computer to perform the following processing: Based on multiple images obtained by the camera device capturing body parts of an authenticator moving toward the camera device in ranging mode, the stopping position of the body parts is predicted. Before the body part stops at the stopping position, the shooting mode of the camera device is switched from the ranging mode to the authentication mode; and The image of the body part captured by the camera device in the authentication mode is used to perform liveness authentication on the person being authenticated.

Images