Image processing device, image processing system, image processing method, and program

Abstract

To provide an image processing device, image processing system, image processing method, and program for performing facial authentication with a simple configuration that minimizes unauthorized authentication breakthrough.SOLUTION: An image processing device 2 of an image processing system disclosed herein comprises a processing unit 26 configured to perform an optical flow using a first facial image and a second facial image included in a first image and a second image of a face of an authentication target captured in a chronological order, and a determination unit 23 configured to determine whether the authentication target is a living body or not on the basis of facial movement vectors MV obtained by the optical flow.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to an image processing apparatus, an image processing system, an image processing method, and a program.

Background Art

[0002] Conventionally, as one of the biometric authentication technologies, a face authentication technology is known in which a face image of a person to be authenticated is acquired and compared with a pre-registered face image to authenticate whether the person is the legitimate user. In order to prevent unauthorized breakthrough of face authentication, for example, the biometric authentication device described in Patent Document 1 irradiates an optical pattern to acquire three-dimensional information of the face, generates a three-dimensional image of the face of the authentication target to perform face authentication, and also performs iris authentication simultaneously.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the biometric authentication device disclosed in Patent Document 1, in order to acquire three-dimensional information of the face and iris information, it is necessary to incorporate a projection device or the like that irradiates two types of optical patterns, and it is also necessary to execute a large amount of various calculations related to the three-dimensional information. Therefore, the configuration of the biometric authentication device becomes complicated.

[0005] The present invention has been made to solve the above problems, and an object thereof is to provide an image processing apparatus, an image processing system, an image processing method, and a program that perform face authentication with a simple configuration capable of suppressing unauthorized authentication breakthrough.

Means for Solving the Problems

[0006] To achieve the above object, the image processing apparatus of the present invention A processing unit that performs optical flow using the first and second facial images contained in the first and second images, which are time-series images of the face to be authenticated, A discrimination unit that determines whether the object to be authenticated is a living organism based on the facial movement vector obtained by the optical flow, The editorial department extracts the first and second facial images from the first and second images, Equipped with 、 The editorial department divided the first face image and the second face image into multiple regions, The discrimination unit determines that the authentication target is a biological entity when the sum of the squared differences of the motion vectors of the plurality of regions is greater than or equal to the operating threshold. ru. [Effects of the Invention]

[0007] According to the present invention, it is possible to perform facial recognition with a simple configuration that can suppress unauthorized authentication bypass. [Brief explanation of the drawing]

[0008] [Figure 1] (a) is a simplified diagram showing a first facial image obtained through biometric authentication using an image processing system according to an embodiment of the present invention, and (b) is a simplified diagram showing a second facial image. [Figure 2] (a) is a simplified diagram showing a first facial image when biometric authentication fails in the image processing system according to an embodiment of the present invention, and (b) is a simplified diagram showing a second facial image. [Figure 3] This figure illustrates the configuration of an image processing system according to an embodiment of the present invention. [Figure 4] This is a block diagram of an image processing system according to an embodiment of the present invention. [Figure 5] This figure shows the data obtained by performing an optical flow using Figures 1(a) and 1(b). [Figure 6] This figure shows the data obtained by performing an optical flow using Figures 2(a) and 2(b). [Figure 7] This is a flowchart of a biometric detection-enabled facial recognition process according to an embodiment of the present invention. [Modes for carrying out the invention]

[0009] An image processing system according to an embodiment of the present invention will be described in detail below with reference to the drawings.

[0010] [Overview of the Image Processing System] The image processing system according to an embodiment of the present invention performs facial recognition and biometric detection to enhance the reliability of facial recognition in a managed area (security area) where a limited number of staff (which may include the person to be authenticated T) are permitted to enter, such as a commercial facility. For biometric detection, for example, a first facial image F1, which is one frame from a video showing the face of the person to be authenticated T as shown in Figure 1(a), and in which the face is divided into multiple regions R, and a second facial image F2, shown in Figure 1(b), which is the frame following the first facial image F1, are used. For facial recognition, the first facial image F1 or the second facial image F2, or the facial image before it is divided into multiple regions R, are used. In the image processing system 1, biometric detection is performed after facial recognition.

[0011] Facial recognition is performed using a visual method, which is a common authentication method. Biometric detection involves detecting the movement of the face of the person to be authenticated T using a first facial image F1 and a second facial image F2, and then performing a process (optical flow) to represent that movement as a vector. The sum of squared differences (SSD) of the movement vectors MV of each region R of the face of the person to be authenticated T obtained by the optical flow is acquired, and if there is a region where the sum of squared differences is greater than or equal to the operating threshold, the person to be authenticated T is determined to be a living being and is permitted to enter or exit. If the sum of squared differences is less than the operating threshold, the person to be authenticated T is determined not to be a living being, and the person to be authenticated T is not permitted to enter or exit. In this way, the image processing system 1 can suppress unauthorized authentication bypass in facial recognition with a simple configuration without requiring any special equipment or configuration.

[0012] [Image Processing System Configuration] The configuration of the image processing system 1 will be described below. As shown in FIG. 3, the image processing system 1 includes an image processing device 2 that performs biometric detection and face authentication, an imaging device 4 that captures an authentication target T, and a display device 6 that displays the captured authentication target T, authentication results, and the like. As shown in FIG. 4, the image processing device 2 includes a control unit 20 that performs optical flow, a storage unit 30 that stores face images, motion vectors, etc. of the authentication target T, a communication unit 50 that communicates with the imaging device 4, the display device 6, and the entry / exit gate, and an input unit 70 that inputs information, etc.

[0013] The control unit 20 is composed of at least one processor such as a CPU (Central Processing Unit), and by executing programs and the like stored in the storage unit 30, it realizes the functions of each part (image acquisition unit 21, face authentication unit 22, discrimination unit 23, editing unit 24, feature point detection unit 25, processing unit 26) described later. In addition, the control unit 20 has a clock (not shown) and can acquire the current date and time, count the elapsed time, etc.

[0014] The storage unit 30 is composed of at least one memory such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and part or all of the ROM is composed of an electrically rewritable memory (such as a flash memory). Functionally, the storage unit 30 has an image storage unit 31, a face authentication image storage unit 32, a face image storage unit 33, a feature point storage unit 34, and a setting storage unit 35. Programs executed by the CPU of the control unit 20 and data necessary in advance for executing the programs are stored in the ROM. Data created or changed during program execution is stored in the RAM.

[0015] The first image, the second image, etc. captured by the imaging unit 40 are stored in the image storage unit 31.

[0016] The face authentication image storage unit 32 stores face images of staff members etc. whose entry into the management target area is permitted and which are used by the face authentication unit 22 during face authentication. When there are changes in the face image, such as when glasses become necessary, glasses become unnecessary by using contacts, the hair is cut short, etc., or when a staff member permitted to enter the management area retires or joins the company, etc., the administrator of the image processing system 1 can register new face images, delete old face images, etc.

[0017] The face image storage unit 33 stores the first face image F1, the second face image F2, etc. edited by the editing unit 24.

[0018] The feature point storage unit 34 stores the feature points detected by the feature point detection unit 25 together with the first image and the second image.

[0019] The setting storage unit 35 stores various settings used in the image processing device 2. Setting values necessary for biometric detection corresponding face authentication processing, such as the number of frames for determining whether the face orientation and angle are the same, biometric detection tolerance difference, operation threshold value, etc. are stored, and the administrator of the image processing system 1 can change these settings via the input unit 70 as needed. Here, the face orientation is what changes by rotating the face left and right around the neck. The face angle represents the angle of the face with the top of the head as the midpoint.

[0020] The communication unit 50 has a communication device 51 which is a module for communicating with external devices such as entry / exit gates. The communication device 51 is a wireless module including an antenna when communicating with external devices. For example, the communication device 51 is a wireless module for performing short-range wireless communication based on Bluetooth (registered trademark). Through the communication unit 50, the image processing device 2 controls the imaging device 4, the display device 6, entry / exit gates, etc. to manage the entry / exit of the authentication target T and the security of the management target area.

[0021] The input unit 70 has an input device 71 for the authentication target T to input various information to the image processing device 2. The input unit 70 also has a keyboard, a mouse, etc. for administrators, the authentication target T, etc. to input information, etc.

[0022] The imaging device 4 includes an imaging unit 40 that images the object to be authenticated T, and a drive unit 41 that changes the orientation of the imaging unit 40.

[0023] The imaging unit 40 is equipped with a CMOS (Complementary Metal Oxide Semiconductor) camera. The imaging unit 40 captures the authentication target T and generates a first image, a second image, etc.

[0024] The drive unit 41 changes the orientation of the imaging unit 40. This changes the imaging range and imaging angle of the imaging unit 40, allowing for appropriate imaging of the face of the subject T being authenticated.

[0025] The display device 6 includes a display unit 60 made up of a liquid crystal display panel (LCD).

[0026] The display unit 60 is composed of a thin-film transistor (TFT), liquid crystal, or organic EL, etc. The display unit 60 displays a first face image F1, a second face image F2, etc.

[0027] Next, the functional configuration of the control unit 20 of the image processing device 2 will be described. The control unit 20 implements the functions of the image acquisition unit 21, face recognition unit 22, discrimination unit 23, editing unit 24, feature point detection unit 25, and processing unit 26, and performs face recognition biometric detection entry / exit processing, etc., which will be described later.

[0028] The image acquisition unit 21 causes the imaging unit 40 to capture an image within a range pre-set in the image processing device 2 or set by the user, and acquires a video including the first image, the second image, etc. The image acquisition unit 21 transmits the acquired first image to the face recognition unit 22. The image acquisition unit 21 also stores the video including the acquired first image and the second image in the image storage unit 31.

[0029] The facial recognition unit 22 performs facial recognition using the first image transmitted from the image acquisition unit 21. The facial recognition unit 22 recognizes the positions of eyes, nose, mouth, etc., in the image and compares them with the facial image stored in the facial recognition image storage unit 32 to authenticate that it is a specific person, performing known 2D facial recognition.

[0030] The discrimination unit 23 determines whether the orientation and angle of the face of the subject T to be authenticated in each image are the same. The discrimination unit 23 also determines whether the difference between the face position information (position of the face in the image) in the first image and the face position information in the second image is within the acceptable difference for biometric detection. Furthermore, it determines whether there is a region R where the sum of the squared differences of the motion vectors MV of each region R, after subtracting the motion vector GMV of the entire face image, is greater than or equal to the operating threshold.

[0031] The editing unit 24 extracts the first face image F1 and the second face image F2, which show the frontal part of the face of the subject T being authenticated and its surroundings, from the first and second images acquired by the image acquisition unit 21, into 224 pixels × 224 pixels. It also adjusts the extracted second face image F2 to the same size as the first face image F1. Furthermore, as shown in Figures 1 and 2(a) and (b), multiple rectangular regions R, each 5 × 5 in this embodiment, are set on the first face image F1 and the second face image F2. The size of each region is 32 pixels × 32 pixels. The editing unit 24 also offsets the starting point of the second face image F2 to adjust the size of the face images and aligns the faces of the first face image F1 and the second face image F2. Details of the starting point offset and face alignment will be described later. The editing unit 24 stores the extracted first face image F1 and second face image F2 in the face image storage unit 33.

[0032] The feature point detection unit 25 detects feature points within multiple regions R of the first face image F1 and the second face image F2. The feature point detection unit 25 stores the detected feature points in the feature point storage unit 34.

[0033] The processing unit 26 performs optical flow analysis on the first face image F1 and the second face image F2 using the feature points detected by the feature point detection unit 25 to obtain motion vectors MV for multiple regions R. It also obtains the motion vector GMV for the entire face image. Furthermore, it calculates the sum of squared differences of the motion vectors MV for each region R after subtracting the motion vector GMV for the entire face image. The Lucas-Kanade method is used to calculate the optical flow.

[0034] The functional configuration of the control unit 20 has been described above. Below, the biometric detection-enabled facial recognition processing performed by the image processing system 1 will be specifically described using the cases where the first facial image F1 is shown in Figure 1(a) and the second facial image F2 is shown in Figure 1(b), and where the first facial image F1 is shown in Figure 2(a) and the second facial image F2 is shown in Figure 2(b) as examples. It is assumed that facial images identical to the facial images F1 and F2 shown in Figures 1(a), 2(a), and (b) (i.e., the facial images shown in the first and second images) are stored in the facial recognition image storage unit 32. It is also assumed that the orientation and angle of the face of the person to be authenticated T remain the same for a specified number of frames. It is assumed that the difference in the facial position of the person to be authenticated T in the first and second images is within the biometric detection tolerance, and that the data exemplified in Figure 5 is obtained as a result of performing optical flow using the facial images F1 and F2 shown in Figures 1(a) and (b). Furthermore, it is assumed that the data shown in Figure 6 was obtained as a result of performing optical flow using facial images F1 and F2 as shown in Figures 2(a) and (b).

[0035] (Assuming the first facial image F1 is Figure 1(a) and the second facial image F2 is Figure 1(b)) As shown in Figure 3, the person to be authenticated T stands in front of the imaging device 4 of the image processing system 1, and the imaging device 4 captures the face of the person to be authenticated T in a video. The image acquisition unit 21 acquires a first image and a second image from the video, and transmits the first image to the face authentication unit 22.

[0036] The facial recognition unit 22 compares the facial image captured in the first image with the facial recognition image storage unit 32. As set in paragraph 0034, since the facial recognition image storage unit 32 stores a facial image identical to the facial image captured in the first image, authentication is successful.

[0037] Furthermore, as set in paragraph 0034, the face of the subject T to be authenticated maintains the same orientation and angle for the specified number of frames, and the difference in the position of the subject T's face in the first and second images is within the acceptable range for biometric detection. Therefore, the editorial department 24 generates the first face image F1 and the second face image F2.

[0038] After adjustment, multiple regions R (5 × 5 = 25 regions R in this embodiment) are set to include the areas where the eyes, nose, and mouth of the face are located, and the first face image F1 and the second face image F2 shown in Figures 1(a) and (b) are obtained. The editing unit 24 offsets the starting point of the second face image F2 in accordance with the size adjustment to align the faces of the first face image F1 and the second face image F2. The editing unit 24 cuts out the first face image F1 from the first image and the aligned second face image F2 from the second image, and adjusts the second face image F2 to be the same size as the first face image F1. The method of offsetting and adjusting the starting point will be described later in step S6 of the whole-body detection-compatible face recognition process. The starting point is the vertex of the upper left corner of the face image.

[0039] The feature point detection unit 25 detects feature points within multiple regions R of the first face image F1 and the second face image F2.

[0040] The processing unit 26 performs optical flow using the detected feature points with the first face image F1 and the second face image F2 to obtain motion vectors MV of multiple regions R. It also obtains motion vector GMV of the entire face image.

[0041] The motion vector MV and the motion vector GMV of the entire face image are obtained as shown in Figure 5, and the sum of the squared differences of the motion vector MV of each region R is calculated by subtracting the motion vector GMV of the entire face image.

[0042] Since there are five regions R where the sum of squared differences exceeds 48 (there are five numbers enclosed in [] in the SSD(-GMV) section of Figure 5), the discrimination unit 23 determines that the authentication target T shown in Figures 1(a) and (b) is a living organism, and determines that facial recognition has been successfully performed, granting the authentication target T permission to enter and exit.

[0043] (Assuming the first facial image F1 is Figure 2(a) and the second facial image F2 is Figure 2(b)) Paragraphs 0035-0037 and 0039-0040 are omitted because the first and second facial images are identical to those in Figures 1(a) and (b). The following explanation will only cover paragraphs where the first and second facial images differ from those in Figures 1(a) and (b).

[0044] The editor 24 extracts the first face image F1 from the first image and the second face image F2 from the second image, and adjusts the second face image F2 to be the same size as the first face image F1. After adjustment, multiple regions R (5 × 5 = 25 regions R in this embodiment) are set to include the areas where the eyes, nose, and mouth of the face are located, and the first face image F1 and the second face image F2 shown in Figures 2(a) and (b) are obtained.

[0045] The motion vector MV and the motion vector GMV of the entire face image are obtained as shown in Figure 6, and the sum of the squared differences of the motion vector MV of each region R is calculated by subtracting the motion vector GMV of the entire face image.

[0046] Since there is no region R where the sum of squared differences exceeds 48 (there are no values ​​enclosed in [] in the SSD(-GMV) items in Figure 6), the discrimination unit 23 determines that the authentication target T shown in Figures 2(a) and (b) is not a biological entity, and determines that an unauthorized bypass of facial recognition has been attempted, resulting in authentication failure, and does not grant entry or exit permission to the authentication target T.

[0047] The following flowchart illustrates the biometric detection-enabled facial recognition process performed by the image processing system 1.

[0048] (Biometric facial recognition processing) Referring to Figure 7, the flow of the biometric detection-enabled facial recognition process is explained. The biometric detection-enabled facial recognition process can suppress unauthorized authentication bypasses of facial recognition. Image processing system 1 performs access control by combining facial recognition and biometric detection. Even if facial recognition is successful, if the subsequent biometric detection does not determine that the person to be authenticated T is a living being, the person to be authenticated T will not be permitted to enter or exit.

[0049] First, the image acquisition unit 21 acquires a first image from the video in which the face of the person to be authenticated T is captured (step S1).

[0050] The process proceeds to step S2, where the face recognition unit 22 performs face recognition using the first image. If a face image that can be considered identical to the face in the first image is found in the images stored in the face recognition image storage unit 32, and face recognition is successful (step S2: Yes), the process proceeds to step S3. If face recognition fails (step S2: No), a non-authentication message is displayed on the display unit 60 (step S14) and the process terminates.

[0051] In step S3, the image acquisition unit 21 acquires a second image, which is a frame consecutive to the first image, from the video in which the face of the person to be authenticated T is captured, and proceeds to step S4.

[0052] In step S4, the discrimination unit 23 determines whether the orientation and angle of the face of the subject to authentication T are the same in the 8 frames immediately preceding the first image and the 2 frames of the first and second images, for a total of 10 frames (hereinafter referred to as face placement confirmation frames). If the orientation and angle of the face of the subject to authentication T are the same in the face placement confirmation frames (step S4: Yes), the process proceeds to step S5. If the orientation and angle of the face of the subject to authentication T are not the same (step S4: No), a decrease in the accuracy of biometric detection is expected, so biometric detection is not performed, and the process returns to step S1.

[0053] In step S5, the discrimination unit 23 determines whether the difference between the face position (coordinate) information in the first image and the face position (coordinate) information in the second image is within 30 pixels, which is the allowable difference for biometric detection. If the difference is within 30 pixels (step S5: Yes), the process proceeds to step S6. If the difference exceeds 30 pixels (step S5: No), a decrease in the accuracy of biometric detection is expected, so biometric detection is not performed, and the process returns to step S1.

[0054] In step S6, the editorial staff 24 compares the sizes of the first face image and the second face image. If the sizes of the first and second face images are different, the editorial staff offsets the starting point of the second face image F2 to align the faces of the first face image F1 and the second face image F2. Specifically, the alignment process involves one of the following two operations: If "size of the first face image > size of the second face image", the editorial staff 24 shifts the coordinates of the starting point of the second face image in the negative direction by ((size of the first face image - size of the second face image) ÷ 2). For example, if the starting point of the first face image is (100, 100) and the size of the first face image is 300 width and 300 height, and the starting point of the second face image is (100, 100) and the size of the second face image is 280 width and 280 height, then the starting point coordinates of the second face image are set to (90, 90) by subtracting (300-280)÷2 from the starting point (100, 100). Then, in step S7, the size of the second face image is enlarged to the size of the first face image. If "size of the first face image < size of the second face image", the coordinates of the starting point of the second face image are shifted in the positive direction by ((size of the second face image - size of the first face image)÷2). For example, if the starting point of the first face image is (100, 100) and the size of the first face image is 260 in width and 260 in height, and the starting point of the second face image is (100, 100) and the size of the second face image is 300 in width and 300 in height, then the starting point of the second face image is (120, 120) by adding (300-260)÷2 each time. Then, in step S7, the size of the second face image is reduced to the size of the first face image.

[0055] In step S7, the editorial team 24 extracts the first face image F1 from the first image and the second face image F2, which has been aligned, from the second image. At this time, in order to reduce the influence of facial movement (slight movement) of the subject T being authenticated and variations in facial information, the editorial team 24 adjusts the size of the second face image F2 to be the same as the size of the first face image F1. In addition, multiple regions R are set in the first face image F1 and the second face image F2.

[0056] The process proceeds to step S8, where the feature point detection unit 25 detects feature points from the first face image F1 and the second face image F2. The processing unit 26 uses the detected feature points and the first face image F1 and the second face image F2 after face alignment to perform optical flow.

[0057] In step S9, the processing unit 26 obtains the motion vector GMV of the entire face image of the subject to authentication T by executing the optical flow in step S8.

[0058] In step S10, the processing unit 26 calculates the motion vector MV of each region R of the face image by subtracting the motion vector GMV of the entire face image of the subject T to be authenticated. Then, the processing unit 26 obtains the sum of the squared differences of the calculated motion vector MV of each region R.

[0059] The process proceeds to step S11, where the discrimination unit 23 determines whether there was a region R where the sum of squared differences was greater than or equal to the operating threshold of 48. If there was a region R where the sum of squared differences was greater than or equal to 8 (step S11: Yes), the authentication target T is determined to be a biological entity, and the facial recognition authentication was deemed successful. Entry and exit permission is then granted (step S12), and the process terminates. If there was no region R where the sum of squared differences was greater than or equal to 48 (step S11: No), the authentication target T is determined to be a biological entity, and an unauthorized attempt to bypass facial recognition authentication was made. Entry and exit permission is not granted (step S13), and the process terminates.

[0060] The above explains the facial recognition process with biometric detection capabilities.

[0061] Thus, with biometric detection-enabled facial recognition processing, the image processing system 1 uses the first and second facial images F1 and F2, which are time-series images of the face of the person to be authenticated T, to obtain the sum of the squared differences of the motion vectors MV of each part of the face in the facial images using optical flow and compare it with an operating threshold to determine whether the person to be authenticated T is a living being. This simple configuration allows for the detection and suppression of fraudulent authentication bypasses of facial recognition using pre-prepared still images, posters, dolls, etc. Furthermore, by extracting the first and second facial images F1 and F2, and dividing each of them into multiple regions R to obtain the motion vectors MV, the computational complexity can be significantly reduced.

[0062] Furthermore, the image processing system 1 can improve the accuracy of the motion vector MV by matching the sizes of the first face image F1 and the second face image F2. In addition, the accuracy of the motion vector MV can be improved by adjusting the sizes of multiple regions R and their motion vectors MV in conjunction with matching the sizes of the first face image F1 and the second face image F2. Moreover, the motion vector MV of multiple regions R can be used to extract the movement of feature points within each region of the face image by subtracting the motion vector of the face of the subject T obtained from the first and second images beforehand.

[0063] (modified version) In the above embodiment, the image processing system 1 determined that the object was a living organism if the sum of squared differences was equal to or greater than the operating threshold. However, the system may also determine that the object to be authenticated, T, is a living organism if the sum of absolute differences (SAD) is equal to or greater than the operating threshold.

[0064] Furthermore, in the above embodiment, the image processing system 1 had an image processing device 2, an imaging device 4, and a display device 6, but these may be replaced by a single device such as a tablet. Alternatively, for example, the display device and the imaging device may be configured as a single device.

[0065] In the above embodiment, the image processing device 2 performed biometric detection and facial recognition, but facial recognition may be performed by an external device.

[0066] In the above embodiment, the operating threshold was set to 48 in the image processing system 1, but it is preferable to set an appropriate threshold according to conditions such as the distance between the authentication target T and the imaging device 4, and the resolution of the imaging device 4. In addition, it is preferable to set appropriate thresholds for the tolerance difference in biometric detection, etc.

[0067] Furthermore, in the above embodiment, the image processing system 1 was divided into a 5x5 region R as shown in Figures 1(a), (b) and 2(a), (b), but the number of divisions may be changed, or for example, a region R near the cheek where there is almost no movement may not be set, or may be excluded from the optical flow. Alternatively, it may be treated as a single region without division, or motion vectors of each part of the face of the authentication target T (e.g., eyes, nose, mouth, eyebrows, facial contour, etc.) may be calculated directly without dividing it into region R.

[0068] In the above embodiment, the image processing system 1 used the second image as the frame following the first image. However, the second image may be two frames after the first image, or even later, as long as the frame rate allows for the detection of the blink of the subject T being authenticated.

[0069] In the above embodiment, the image processing system 1 performed biometric detection after facial recognition, but facial recognition may be performed after biometric detection, or biometric detection and facial recognition may be performed in parallel.

[0070] In the above embodiment, the image processing system 1 performed 2D facial recognition, but other facial recognition methods, such as facial recognition using an infrared sensor, may also be used.

[0071] In the above embodiment, the image processing system 1 was used for access control, but it may also be used for attendance management, PC security, logon authentication for tablets and smartphones, etc., or it may be incorporated into robots or lockers.

[0072] In the above embodiment, the image processing system 1 cropped the first face image F1 and the second face image F2 to 224 pixels × 224 pixels and edited the 32 pixels × 32 pixels region R into a 5 × 5 rectangle. However, these values ​​may be adjusted as appropriate.

[0073] Each function of the image processing system 1 of this invention can also be performed by a regular PC (Personal Computer) or tablet. Specifically, in the above embodiment, the program for biometric detection-compatible facial recognition processing performed by the image processing system 1 was described as being pre-stored in the ROM of the storage unit 30. However, a computer capable of realizing the above functions may be configured by distributing the program on a computer-readable recording medium such as a flexible disk, CD-ROM (Compact Disc Read Only Memory), DVD (Digital Versatile Disc), and MO (Magneto-Optical Disc), and then loading and installing the program into the computer.

[0074] Although preferred embodiments of the present invention have been described above, the present invention is not limited to these specific embodiments, and the present invention includes the invention described in the claims and its equivalents. The invention described in the original claims of this application is listed below.

[0075] (Note) (Note 1) A processing unit that performs optical flow using the first and second facial images contained in the first and second images, which are time-series images of the face to be authenticated, A discrimination unit that determines whether the object to be authenticated is a living organism based on the facial movement vector obtained by the optical flow, An image processing device equipped with the following features.

[0076] (Note 2) The system further includes an editing unit that extracts the first face image and the second face image from the first image and the second image. The image processing apparatus described in Appendix 1.

[0077] (Note 3) The editorial department divided the first face image and the second face image into multiple regions, The discrimination unit determines whether or not the object to be authenticated is a living organism based on the motion vectors of the plurality of regions. The image processing apparatus described in Appendix 2.

[0078] (Note 4) If the sum of the squared differences of the motion vectors of the aforementioned multiple regions is greater than or equal to the operating threshold, it is determined that the object to be authenticated is a living organism. The image processing apparatus described in Appendix 3.

[0079] (Note 5) The editorial department will make the size of the first face image and the second face image the same. The image processing apparatus described in Appendix 4.

[0080] (Note 6) In order to make the sizes of the first face image and the second face image the same, the size and motion vector of the multiple regions are adjusted. The image processing apparatus described in Appendix 5.

[0081] (Note 7) The motion vectors of the aforementioned multiple regions have the motion vectors of the face to be authenticated, obtained from the first and second images, subtracted in advance. The image processing apparatus described in Appendix 6.

[0082] (Note 8) An image processing device described in any one of the appendices 1 to 7, An imaging device for capturing images of the object to be authenticated, A display device that presents the authentication result to the object to be authenticated, An image processing system having the following features.

[0083] (Note 9) An optical flow is performed using the first and second facial images contained in the first and second images, which capture the face to be authenticated in chronological order. Based on the facial movement vector obtained by the optical flow, it is determined whether or not the object to be authenticated is a living organism. Image processing methods.

[0084] (Note 10) On the computer, An optical flow is performed using the first and second facial images contained in the first and second images, which capture the face to be authenticated in chronological order. Based on the facial movement vector obtained by the optical flow, it is determined whether or not the object to be authenticated is a living organism. A program designed to make it work in that way. [Explanation of Symbols]

[0085] 1…Image processing system, 2…Image processing device, 4…Imaging device, 6…Display device, 20…Control unit, 21…Image acquisition unit, 22…Face recognition unit, 23…Discrimination unit, 24…Editing unit, 25…Feature point detection unit, 26…Processing unit, 30…Storage unit, 31…Image storage unit, 32…Face recognition image storage unit, 33…Face image storage unit, 34…Feature point storage unit, 35…Setting storage unit, 40…Imaging unit, 41…Drive unit, 50…Communication unit, 51…Communication device, 60…Display unit, 70…Input unit, 71…Input device, F1…First face image, F2…Second face image, GMV…Motion vector of the entire face image, MV…Motion vector, R…Region, T…Authentication target

Claims

1. A processing unit that performs optical flow using a first face image and a second face image contained in a first image and a second image of the face to be authenticated captured in chronological order, A discrimination unit that determines whether the object to be authenticated is a living organism based on the facial movement vector obtained by the optical flow, The editorial department extracts the first face image and the second face image from the first image and the second image, Equipped with, The editorial department divides the first face image and the second face image into multiple regions, The discrimination unit determines that the object to be authenticated is a living organism when the sum of the squared differences of the motion vectors of the plurality of regions is greater than or equal to the operating threshold. Image processing device.

2. The editorial department will make the size of the first face image and the second face image the same. The image processing apparatus according to claim 1.

3. In order to make the sizes of the first face image and the second face image the same, the size and motion vector of the multiple regions are adjusted. The image processing apparatus according to claim 2.

4. The motion vectors of the aforementioned multiple regions have the motion vectors of the face to be authenticated, obtained from the first image and the second image, subtracted in advance. The image processing apparatus according to claim 3.

5. An image processing apparatus according to any one of claims 1 to 4, An imaging device for capturing images of the object to be authenticated, A display device that presents the authentication result to the object to be authenticated, An image processing system having the following features.

6. An image processing method performed by an image processing device, A processing step of performing optical flow using a first face image and a second face image contained in a first image and a second image of the face to be authenticated captured in chronological order, A determination step of determining whether the subject to authentication is a living organism based on the facial movement vector obtained by the optical flow, An editing step of extracting the first face image and the second face image from the first image and the second image, It has, In the editing step described above, the first face image and the second face image are each divided into multiple regions, In the aforementioned determination step, if the sum of the squared differences of the motion vectors of the multiple regions is greater than or equal to the operating threshold, it is determined that the object to be authenticated is a living organism. Image processing methods.

7. Computers, Processing means for performing optical flow using a first face image and a second face image contained in a first image and a second image of the face to be authenticated, captured in chronological order. A determination means for determining whether the object to be authenticated is a living organism based on the facial movement vector obtained by the optical flow, Editing means for extracting the first face image and the second face image from the first image and the second image, To make it function as, The editing means divides the first face image and the second face image into multiple regions, The determination means determines that the object to be authenticated is a living organism when the sum of the squared differences of the motion vectors of the plurality of regions is greater than or equal to the operating threshold. program.

Images