Face Recognition System

Abstract

To provide a face authentication system robust to individual differences in facial parts or changes in face orientation when identifying an individual by using a face image of a person wearing a wearable item.SOLUTION: The face authentication system deforms an image of a wearable item to match the face shape and face orientation of the registered user's real face image when a wearable item such as a mask, sunglasses, goggles, eyeglasses, a hat, etc. are detected from an image of a user's face at the time of authentication to fit a real face image. The user's real face image is matched against the user's authentication image using an image that combines the bare-face image and the wearable item image.SELECTED DRAWING: Figure 11

Description

[Technical field]

[0001] The present invention relates to a face authentication system that identifies an individual using a face image. [Background technology]

[0002] Facial recognition technology, one of the biometric authentication technologies that uses biometric information, is a technology that registers the facial image of an individual user in advance and determines whether the person is the person in question by comparing the facial image extracted during authentication with the pre-registered facial image. Facial recognition technology has the advantage that it can perform authentication without contact because it can capture facial images from a distance, and that, unlike fingerprint authentication, the user does not need to perform any authentication action. The scope of use of facial recognition is expanding, including entrance and exit management in offices, educational institutions, event venues, etc., and surveillance and security applications using surveillance camera footage.

[0003] In particular, in recent years, due to the impact of COVID-19, there has been an increased need to verify identities while wearing masks and goggles, and there is a demand for systems that can perform facial recognition even when wearing protective gear such as masks.

[0004] In response to the above-mentioned problems, for example, Patent Document 1 discloses a technology that uses an image in which an image of an object such as a mask or hat is combined with an image of an unmasked face as the registered face image used during matching, thereby reducing the burden on the user during registration and performing face authentication while the object is still being worn. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2015-088095 A Summary of the Invention [Problem to be solved by the invention]

[0006] The method described in Patent Document 1 synthesizes an attached image directly onto a face image, so it is not possible to generate an attached image that corresponds to individual differences in facial parts (arrangement, size, shape, etc.) and changes in face direction (rotation in the yaw, roll, and pitch angles), and is therefore less robust against changes in face state. In addition, in applications such as simultaneous authentication of multiple people using signage and face authentication using surveillance cameras, the face direction at the time of authentication is not limited to the front face, but is expected to be facing in various directions. Therefore, a face authentication system that is highly robust against face direction is desired.

[0007] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a face recognition system that is robust against individual differences in facial features or fluctuations in facial direction when identifying an individual using a facial image of the individual wearing an attachment. [Means for solving the problem]

[0008] The face authentication system according to the present invention identifies an individual by using a composite image obtained by modifying an attached image so as to match the individual's facial shape. Effect of the Invention

[0009] According to the face recognition system of the present invention, when identifying an individual using a face image of the individual wearing an accessory, it is possible to improve robustness against individual differences in facial features or fluctuations in facial direction, thereby providing a highly accurate face recognition system. [Brief description of the drawings]

[0010] [Figure 1] 1 is a block diagram showing the overall configuration of a face authentication system 1 according to a first embodiment. [Figure 2A] This is an example of a natural face image 401. [Figure 2B] 4 is an example of an attachment image 402. [Figure 2C] 4 is an example of a composite image 403. [Diagram 3]2 is a block diagram showing the internal configuration of a face image registration unit 220 and a face image matching unit 320. FIG. [Figure 4] 11 is a flowchart for explaining a process of registering an unmasked face image 401 during face registration. [Diagram 5] FIG. 4 is a diagram showing a process for generating a composite image 403 from a natural face image 401. [Figure 6] 13 is a flowchart for explaining a process of generating a composite image 403 from a natural face image 401 and recording it in a data storage unit 400. [Figure 7] 11 is a flowchart illustrating a process during face authentication. [Figure 8A] 13A and 13B are diagrams showing the results of comparing similarities when the face is tilted in the yaw direction from the front. [Figure 8B] 13A and 13B are diagrams showing the results of comparing similarities when the face is tilted in the roll direction from the front. [Figure 9] FIG. 11 is a diagram showing a detailed configuration of a face authentication system 1 in a second embodiment. [Figure 10] 11 is a flowchart illustrating a process for face registration in the second embodiment. [Figure 11] 13 is a flowchart illustrating a process for face authentication in the second embodiment. [Figure 12] FIG. 11 is a configuration diagram of a face authentication system 1 according to a third embodiment. [Figure 13] FIG. 2 is a diagram showing internal processing blocks of the face authentication system 1. [Figure 14] 11 is a flowchart illustrating a process for generating and registering a composite image 403 during face registration. [Figure 15] FIG. 11 is a diagram illustrating the configuration of a face authentication unit 300 according to a third embodiment. [Figure 16] 13 is a flowchart showing an operation during face authentication in the third embodiment. [Figure 17] FIG. 13 is a diagram showing the configuration of a face authentication unit 300 according to a fourth embodiment. [Figure 18] 13 is a flowchart for explaining an operation upon authentication in the fourth embodiment. [Figure 19]FIG. 13 is a diagram illustrating the configuration of a face registration unit 200 according to a fifth embodiment. [Figure 20] FIG. 13 is a diagram showing the effect of the fifth embodiment. [Figure 21] FIG. 13 is a diagram showing the configuration of a face authentication unit 300 according to a fifth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] <Embodiment 1> 1 is a block diagram showing the overall configuration of a face authentication system 1 according to a first embodiment of the present invention. The face authentication system 1 is a system for identifying an individual using a face image. The face authentication system 1 includes an imaging unit 100, a face registration unit 200, a face authentication unit 300, and a data storage unit 400.

[0012] The face registration unit 200 includes a face area detection unit 210 and a face image registration unit 220. The face authentication unit 300 includes a face area detection unit 310 and a face image matching unit 320. The data storage unit 400 stores an unmasked face image 401, an attachment image 402, and a composite image 403. Both the face registration unit 200 and the face authentication unit 300 may generate the composite image 403 by combining the unmasked face image 401 and the attachment image 402 as described below, and therefore can serve as a "compositing unit" that generates the composite image 403.

[0013] The imaging section 100 may be any image capturing section capable of acquiring a two-dimensional image in which luminance information is stored in the x and y directions, or a three-dimensional image in which distance (depth) information in the z direction is stored in addition to luminance information in the x and y directions.

[0014] The composite image 403 is obtained by processing the accessory image 402 so as to conform to the bare face image 401 and synthesizing it with the bare face image 401. Examples of processing will be described later. The composite image 403 may be stored for each type of external accessory, such as a mask, glasses, sunglasses, goggles, and a hat. If there are multiple shape patterns for the same type of accessory, the composite image 403 may be generated for each shape pattern. For example, if the accessory is a mask, it may be a flat mask or a three-dimensional mask, and if it is glasses, the lens shape may be a square type or a round type. In addition, multiple types of external accessories may be synthesized at the same time. For example, a mask and goggles may be synthesized. In the present invention, the bare face image 401, the accessory image 402, and the composite image 403 may be either a two-dimensional image or a three-dimensional image, and may be a feature vector amount instead of an image form.

[0015] The face authentication system 1 may have two or more data storage units 400, and may use different data storage units 400 depending on the type of data. For example, the bare face image 401, the attachment image 402, and the composite image 403 may be registered in different external data storage units 400, and may be received wirelessly from the data base station 2.

[0016] The operation during face registration will be described. The captured image captured by the imaging unit 100 is sent to the face area detection unit 210 in the face registration unit 200. The face area detection unit 210 extracts a bare face image from the captured image and inputs it to the face image registration unit 220. The face image registration unit 220 records the input bare face image in the data storage unit 400 as registered face data of the user. The registered face data refers to the bare face image 401 and the composite image 403, and only one of them may be recorded or both may be recorded at the same time. A plurality of types of composite images 403 may be recorded at the same time depending on the type of attachment.

[0017] The operation during face authentication will be described. The captured image captured by the imaging unit 100 is sent to the face area detection unit 310 in the face authentication unit 300. The face area detection unit 310 extracts a face image from the captured image and inputs it to the face image matching unit 320. The face image extracted by the face area detection unit 310 may be either a natural face or a face image when wearing an attachment, depending on the state of the user's face at the time of authentication. The face image matching unit 320 acquires one or more of the natural face image 401 or the composite image 403 from the data storage unit 400. The face image matching unit 320 performs a face image matching process between the acquired registered face data and each of the face images extracted from the captured image at the time of authentication, and outputs the authentication result by matching to an output device or another system. Specifically, the face image matching unit 320 calculates the similarity between the face image extracted by the face area detection unit 310 and each of one or more registered face data acquired from the data storage unit, and if any of the similarities reaches a prescribed level, authenticates the user as a registered user.

[0018] 2A is an example of a bare face image 401. The bare face image 401 is a face image in a state where no external accessories are attached, and is considered to be a bare face image even if the subject is wearing makeup or the like.

[0019] 2B is an example of an accessory image 402. The accessory image 402 is an image of an accessory alone. The accessory here refers to an external artificial object such as a mask, sunglasses, goggles, glasses, hat, etc. that is attached to the bare face (head of the person to be authenticated) and is an article that hinders face authentication.

[0020] 2C is an example of a composite image 403. The composite image 403 is obtained by processing the accessory image 402 to match the facial features and the facial orientation, and synthesizing it to fit the bare face image 401. For example, if the bare face image 401 is rotated around the yaw axis, the accessory image 402 is similarly rotated around the yaw axis and synthesized with the bare face image 401 to generate the composite image 403.

[0021] This embodiment can be applied to both 1:1 authentication, in which the input biometric information and the biometric information to be compared are in a one-to-one relationship, and 1:N authentication, in which the input biometric information and the biometric information to be compared are in a one-to-N relationship. In 1:1 authentication, a card and biometric information are used as an example, and the input biometric information is compared with the biometric information registered on the card. In 1:N authentication, the input biometric information is compared sequentially with all biometric information registered in the database, and the closest user is uniquely identified.

[0022] 3 is a block diagram showing the internal configuration of face image registration unit 220 and face image matching unit 320. Face image registration unit 220 includes face landmark detection unit 221, attachment image processing unit 222, and authentication data generation unit 223. Face image matching unit 320 includes authentication data generation unit 321. The operations of these functional units will be described later.

[0023] FIG. 4 is a flowchart for explaining the process of registering an unmasked face image 401 at the time of face registration. The captured image captured by the imaging unit 100 is input to the face area detection unit 210 (S401). The face area detection unit 210 extracts the face area of ​​the user to be registered from the captured image to generate an unmasked face image (S402). The generated unmasked face image is sent to the authentication data generation unit 223. The authentication data generation unit 223 generates authentication data based on the unmasked face image (S403) and records it in the data storage unit 400 (S404). The authentication data may be the composite image 403 itself, or its feature amount may be used as the authentication data. In the former case, the composite image 403 is compared with the captured image, and in the latter case, the feature amount of the composite image 403 is compared with the feature amount of the captured image.

[0024] In S402, an image area corresponding to a face area is identified from brightness information (or distance information or a combination thereof) in the captured image, and the image area is cut out from the captured image to generate a face image. In S402, when generating a face image from the face area cut out from the captured image, one or more processing steps may be included to process the face image into a state more suitable for face recognition. For example, image size conversion, image rotation, affine transformation, face direction angle deformation by a non-rigid deformation method based on a Free Form Deformation (FFD) method or the like, noise and blur removal contained in the face image, contrast adjustment, distortion correction, background area removal, etc. may be used.

[0025] The face region detection process in S402 can be performed using a known method, such as the Viola-Jopnes method using Haar-like features or a detector based on a Neural Network (NN) model.

[0026] The authentication data generated in S403 is converted into a format suitable for the algorithm adopted by the face authentication unit 300, and may be a feature vector amount or an image form. An example of generating a feature vector will be described below. The authentication data generating unit 223 generates a feature vector representing the features of the face image by inputting the received bare face image to a predetermined feature extractor. As an example of the feature extractor, one based on an NN model trained by learning data, or an algorithm based on Local Binary Patterns, etc., can be used. A plurality of registered face images may be input to the feature extractor at the same time to generate an integrated feature vector amount. For example, a combination of the bare face image 401 and a plurality of types of composite images 403, etc., is input to the feature extractor at the same time to generate a feature vector. The feature vector amount may be obtained by multiplying a predetermined weight by the feature vectors corresponding to each of the plurality of registered face images and adding them together.

[0027] In S404, when the data is recorded in the data storage unit 400, encryption processing or the like may be performed to improve security.

[0028] FIG. 5 is a diagram showing a process of generating a composite image 403 from an unmasked face image 401. The facial landmark detection unit 221 detects facial landmarks 401-1 from the unmasked face image 401. A facial landmark is a reference point on a facial image. The accessory image processing unit 222 generates an accessory image 402-1 after processing by transforming the accessory image 402 so that it fits the position of the facial landmark 401-1. The authentication data generation unit 223 generates a composite image 403 by synthesizing the unmasked face image 401 and the accessory image 402-1 after processing. The synthesis procedure is as follows.

[0029] (Synthesis procedure: Part 1) The face landmark detection unit 221 detects the rotation angle of the face (a combination of one or more of yaw, roll, and pitch) based on the face landmark. The attachment image processing unit 222 rotates the attachment image 402 so that it matches the rotation angle of the face.

[0030] (Synthesis Procedure: Part 2) The wearing image processing unit 222 transforms the wearing image 402 so that the vertical size of the wearing image 402 matches the vertical size of the face landmark at which the wearing image is worn. For example, the distance between the top of the mask (the most protruding part of the part covering the nose) and the bottom of the mask can be set as the vertical size of the wearing image 402. For example, the distance between the third nose landmark from the top and the point located at the tip of the chin landmark can be set as the vertical size of the wearing position. The wearing image 402 is aligned so that both sizes match. If necessary, an affine transformation, a homography transformation, or a combination of them may be used.

[0031] (Synthesis Procedure: Part 3) The wearing image processing unit 222 transforms the wearing image 402 so that the horizontal size of the wearing image 402 matches the horizontal size of the facial landmark at which the wearing image is worn. For example, the distance between the left and right ends of the mask can be set as the horizontal size of the wearing image 402. For example, the distance between the facial landmarks at both the left and right ends can be set as the horizontal size of the wearing position. The wearing image 402 is aligned so that both sizes match. If necessary, an affine transformation, a homography transformation, or a combination of these may be used.

[0032] (Synthesis Procedure: Part 4) The wearing image processing unit 222 may perform the above steps 2 and 3 by dividing the wearing image 402 into a plurality of regions. For example, the steps may be performed for the right and left halves. That is, the wearing image 402 may be deformed so that the vertical size of the wearing image 402 and the vertical size of the wearing position are matched for each of the left and right halves of the wearing image 402, and the horizontal size of the wearing image 402 and the horizontal size of the wearing position are matched for each of the left and right halves of the wearing image 402. The horizontal size in this case may be the distance between a line connecting the third point from the top of the nose landmark and the point located at the tip of the chin landmark, and the right end (or left end) point of the face landmark. For example, when the image is not symmetrical with respect to the center of rotation, such as the processed wearing image 402-1, it is useful to perform individual deformations on the left and right halves of the wearing image 402 in this way.

[0033] (Synthesis Procedure: Supplementary Note 1) The above is an example of a case where the accessory is a mask, but the above procedure can be used for other accessories as well. That is, the rotation angle of the face and the rotation angle of the accessory are matched, and then the accessory image 402 is transformed so that the vertical and horizontal sizes of the accessory and the wearing position match. The wearing position can be determined appropriately for each type of accessory. The same procedure can be used for rotations other than yaw rotation.

[0034] (Synthesis Procedure: Supplementary Note 2) The above is an example of rotating the accessory image 402 to match the rotation angle of the face, and then synthesizing the image so that the horizontal and vertical sizes of the accessory and the position where it is worn match. However, the procedure may be reversed so that the horizontal and vertical sizes of the accessory are determined in advance according to the position where it is worn, and then the rotation process is performed before synthesizing.

[0035] In addition to the above synthesis procedure example, a synthesis procedure may be used in which multiple control points are provided inside the attachment image 402, and the attachment image is deformed nonlinearly so that the control points match the reference points of the attachment position determined based on the face landmarks. The above synthesis procedure example may be combined as necessary.

[0036] Fig. 6 is a flowchart for explaining the process of generating a composite image 403 from an unmasked face image 401 and recording the composite image 403 in the data storage unit 400. In Fig. 6, steps having the same functions as those in Fig. 4 are given the same reference numerals, and detailed explanations thereof will be omitted. Here, an example will be explained in which the attachment image 402 is a mask. Compared to Fig. 4, the flowchart in Fig. 6 adds S601, S602, and S603. These steps will be explained below.

[0037] (Figure 6: Step S601) The bare face image 401 extracted in S402 is input to the face landmark detection unit 221. The face landmark detection unit 221 detects face landmarks 401-1 from the face region. The face landmarks 401-1 are feature points that indicate the positions of each part of the face, such as the eyes, eyebrows, nose, mouth, chin, and contour. The face landmarks 401-1 in FIG. 5 show an example in which feature points of 68 face parts are detected, but if the position of at least one face part is detected, it can be used as the face landmark 401-1. A known method can be used for detecting face landmarks. For example, a method using pattern matching based on an Active Appearance Model, a feature point extraction method based on Regression Trees, or a detector based on a NN model. The bare face image 401 generated in S402 and information on the face landmarks 401-1 detected in S601 are input to the attachment image processing unit 222.

[0038] (Figure 6: Step S602) The attachment image processing unit 222 acquires the attachment image 402 stored in advance in the data storage unit 400, and processes the attachment image 402 based on the face landmarks 401-1 to generate a post-processing attachment image 402-1.

[0039] (Figure 6: Step S603) The authentication data generating unit 223 generates a composite image 403 by combining the processed attachment image 402-1 with the natural face image 401 so that the combined image 403 fits the natural face image 401, and stores the combined image 403 in the data storage unit 400.

[0040] In the above example, one wearing image 402 is acquired in S602, but multiple images may be acquired depending on the type of wearing image, and composite images 403 may be generated for the number of types of wearing image and for the number of shape patterns of the same type of wearing. Also, a composite image 403 in which multiple types of wearing items are worn at the same time may be generated, for example, when a mask and goggles are simultaneously synthesized with an unmasked face image 401.

[0041] In the above method, the composite image 403 is automatically generated from the bare face image 401, so the user only needs to register the bare face image 401, and there is no need to take an image of the user wearing the accessory, which reduces the burden on the user during registration. In addition, during face authentication, by matching the face image extracted from the face area detection unit 310 with the bare face image 401 and the composite image 403 stored in the data storage unit 400, face authentication is possible even if the user is wearing the accessory during face authentication. The composite image 403 is generated by processing and synthesizing the accessory image 402 in accordance with the arrangement of the face parts in the bare face image 401 registered by the user, the size of each face part, the shape of each face part, the shape of the face outline, etc., so that the accuracy of deriving the similarity is high regardless of the individual differences in the face parts of each user and the face orientation such as the rotation angles in the yaw, roll, and pitch directions, and therefore face authentication that is robust to changes in the face state and highly accurate is possible.

[0042] Fig. 7 is a flowchart for explaining the process during face authentication. Compared to Fig. 4, S701 to S704 have been added. The following mainly explains the steps added to Fig. 4.

[0043] (FIG. 7: Steps S401 to S403) During face authentication, as in face registration, the face image generated by face area detection unit 310 is input to authentication data generation unit 321 in face image matching unit 320, and authentication data generation unit 321 generates authentication data.

[0044] (FIG. 7: Steps S701 to S703) The face authentication unit 300 acquires from the data storage unit 400 the original face image 401 and the composite image 403 to be compared with the face image acquired during face authentication (S701). At this time, the face authentication unit 300 checks whether the authentication data of the face image acquired during face authentication and the authentication data acquired from the data storage unit 400 are all present (S702). If the authentication data is not present, the process returns to S701 again. If the authentication data is present, the face image acquired during authentication is compared with each of the multiple registered face data acquired from the data storage unit 400 by calculating the similarity between the two (S703).

[0045] (Figure 7: Step S702: Supplement) Having all the authentication data means that the face recognition system 1 has already generated authentication data for all persons who are assumed to be subjects to be authenticated.

[0046] (Figure 7: Step S703: Supplementary Note 1) As an example of similarity calculation, when the authentication data is a feature vector amount, the well-known Euclidean Norm or Cosine Similarity can be used. In calculating the similarity of a pixel set of a two-dimensional image or a three-dimensional image, a method based on well-known block matching or a method based on a luminance distribution vector of an image can be used. In authentication based on the feature vector amount, the amount of calculation is generally reduced compared to when a pixel set is used, so when it is desired to speed up the matching process, similarity calculation using the feature vector amount is effective.

[0047] (Figure 7: Step S703: Supplementary Note 2) In this step, two or more of the feature vector amount, the two-dimensional image, and the three-dimensional image may be combined to calculate the similarity in an overall manner. For example, when the feature vector amount and the two-dimensional image are used, a predetermined similarity may be calculated for each, and the final sum of the calculated similarity may be determined as the similarity based on a predetermined weighting.

[0048] (FIG. 7: Step S704) The face authentication unit 300 performs face authentication by checking whether the calculated similarity satisfies a predetermined threshold condition. The face authentication unit 300 outputs the authentication result to an output device or another system. For example, the following authentication methods can be used. In the case of 1:1 authentication, if the calculated similarity satisfies a predetermined threshold condition, the user is determined to be a registered user. If the threshold condition is not satisfied, it is determined that the user is a "non-identified user." In the case of 1:N authentication, the registered user with the largest calculated similarity and satisfying the predetermined threshold condition is determined to be the user. If the threshold condition is not satisfied, it is determined that the user is "unregistered."

[0049] 8A to 8B show the results of comparing a case where authentication is performed using a composite image 403 generated by processing an attached image 402 based on the face landmarks of the present invention and fitting it to a bare face image 401, and a case where an attached image 402 is not processed and directly synthesized with a bare face image 401 for authentication. Here, the attached image is used as a mask, and the comparison is performed for the same person. At the time of registration, a bare face image facing a specific angle is registered, and a mask is synthesized based on a predetermined method. At the time of authentication, a real mask is worn and facing a specific angle, and the similarity with the face image after the mask synthesis is calculated. For the authentication data, a feature vector generated by a feature extractor based on a NN model is used, and the Euclidean Norm is used for calculating the similarity. The vertical axis shows the relative similarity for each face angle, when the similarity when the person is facing forward and authenticated is set to 0. The smaller the similarity, the more accurate the detection.

[0050] FIG. 8A shows the results of comparing similarities when the face is tilted in the yaw direction from the front, and FIG. 8B shows the results of comparing similarities when the face is tilted in the roll direction from the front. The dotted line shows the results without processing the attached image, and the solid line shows the results with the attached image processing of the present invention. The similarity is smaller when the composite image 403 generated based on the face landmark is used for authentication than when the attached image is not processed. In particular, looking at the graph for the angle in the roll direction, the relative difference in similarity is about 0.05, and the present invention tends to be effective especially in the roll direction. On the other hand, when the attached image 402 is synthesized without processing it according to the facial parts and direction, the difference in similarity tends to increase as the angle of the face direction increases, and if a certain threshold is set, this has a negative effect on authentication accuracy (for example, when the threshold of similarity is set to 0.15, in both the yaw direction and the roll direction, when the face is rotated from the front to a certain extent, the similarity exceeds the threshold, resulting in "non-identity"). Therefore, it can be seen that the present invention can reduce the effect on derivation of similarity even when the face angle changes, and improves robustness against face direction angles.

[0051] <Embodiment 2> In the second embodiment of the present invention, an example configuration is described in which a pattern of an accessory is detected to determine whether or not the accessory is being worn, and if the accessory is being worn, an accessory image 402 corresponding to the pattern is synthesized.

[0052] Here, the shape pattern of the wearable item includes not only the type of wearable item (mask, glasses, goggles, etc.), but also different shapes of the same type of wearable item (for example, in the case of masks, flat masks, three-dimensional masks, masks for children, or masks for women, etc.).

[0053] Fig. 9 is a diagram showing a detailed configuration of a face authentication system 1 in the present embodiment 2. Compared to the embodiment 1 shown in Fig. 3, the internal configurations of the face image registration unit 220 and the face image matching unit 320 are different. The rest is the same as the embodiment 1.

[0054] The facial image registration unit 220 has an internal block including a facial image recording unit 224. The facial image matching unit 320 includes an authentication data generation unit 321, an accessory shape pattern detection unit 322, an accessory image acquisition unit 323, a face landmark detection unit 324, and an accessory image processing unit 325. Descriptions of blocks that share processing with the first embodiment will be omitted, and only the internal processing of the facial image registration unit 220 and the facial image matching unit 320 will be described.

[0055] FIG. 10 is a flowchart for explaining the process of face registration in the second embodiment. Steps similar to those in the first embodiment are given the same reference numerals and will not be described. In S1001, the bare face image generated by the face area detection unit 210 is sent to the face image recording unit 224 in the face image registration unit 220 and registered in the data storage unit 400. At this time, the bare face image to be registered may be a two-dimensional image or a three-dimensional image. If a feature vector amount is required for authentication, S403 may be added after S402 to generate a feature vector using a feature extractor, or may be recorded simultaneously with the data in the image form described above. When recording in the data storage unit 400 as an image form, the face image to be registered may be subjected to lossless compression processing before being recorded in order to reduce the data capacity.

[0056] 11 is a flowchart for explaining the process during face authentication in the present embodiment 2. Steps similar to those explained in the embodiment 1 are given the same step numbers and explanations are omitted. The following mainly explains the differences in the present embodiment 2.

[0057] The face image generated by the face area detection unit 310 in S402 is sent to the accessory shape pattern detection unit 322. In S1101, the accessory shape pattern detection unit 322 detects the shape of the accessory from the face image. In S1102, the accessory shape pattern detection unit 322 checks the presence or absence of an accessory based on the detection result. As an example of pattern detection, a detector based on known geometric shape pattern matching or a NN model, or a combination thereof, can be used. For example, in the case of geometric shape pattern matching, it can be determined that no accessory is present if there is no matching pattern. The detection method is not limited to this as long as the shape and presence or absence of the accessory can be detected.

[0058] After S1101, the face image extracted from the captured image is sent to the authentication data generation unit 321, and the authentication data generation unit 321 generates authentication data (S403).

[0059] If it is determined in S1102 that the wearable item is worn, the wearable item image acquisition unit 323 selects the wearable item image 402 that best matches the detected shape pattern and acquires it from the data storage unit 400, and also acquires the bare face image 401 registered in advance in S1103. The acquired image is sent to the face landmark detection unit 324, which detects face landmarks 401-1 based on the bare face image 401 (S601). Based on the detected face landmarks 401-1, the acquired wearable item image 402 is processed (S602), and a composite image 403 is generated so as to fit the bare face image 401 (S603). The composite image 403 generated by the wearable item image processing unit 325 is sent to the authentication data generation unit 321, and authentication data is generated (S403). Identity authentication is performed by collating authentication data based on the face image extracted from the captured image at the time of authentication with authentication data based on the generated composite image 403.

[0060] If it is determined in S1102 that the accessory is not being worn at the time of authentication, the accessory image acquisition unit 323 does not acquire the accessory image 402, but acquires only the registered face image 401 (S1104), and the authentication data generation unit 321 generates authentication data for only the face image 401 (S403). By providing S1102, it is possible to skip S601 to S603, and the matching can be performed at a high speed.

[0061] Although not shown in the drawings, instead of acquiring the wear image 402 that best matches the shape pattern of the wear, the wear image acquisition unit 323 may generate a wear image having a shape pattern similar to the detected shape pattern of the wear and use it as the wear image 402. For example, if the user is wearing a mask at the time of authentication, a mask having a shape similar to the detected mask shape may be generated and used as the wear image 402. In this way, it is possible to deal with wearers of any shape, instead of preparing a large number of wear images 402 in the data storage unit 400 in advance. The wear image 402 generated or extracted at this time may be registered as it is in the data storage unit 400 so that it can be used when generating another composite image 403.

[0062] <Embodiment 2: Summary> In the face authentication system 1 according to the second embodiment, the accessory shape pattern detection unit 322 detects the shape pattern of the accessory of the person to be authenticated, and obtains only the accessory image corresponding to the shape pattern from the data storage unit 400 to generate a composite image. This reduces the number of composite images used during face authentication, thereby speeding up the authentication process. This is particularly useful in cases where the range of subjects to be authenticated is specified in advance, such as in 1:1 authentication. On the other hand, in cases where personal identification is performed on an unspecified number of people, such as in 1:N authentication, the right half of FIG. 11 is performed on all subjects to be authenticated, which may result in an excessive computation load. Therefore, the second embodiment is particularly useful in the former case.

[0063] <Embodiment 3> 12 is a configuration diagram of a face recognition system 1 according to a third embodiment of the present invention. In the third embodiment, in addition to the configuration described in the first and second embodiments, a three-dimensional information imaging unit 110 is newly provided. The three-dimensional information imaging unit 110 acquires distance (depth) information (i.e., the distance from the three-dimensional information imaging unit 110 to the imaging location) in addition to luminance information of the subject. By using these, it is possible to process and synthesize the attachment more faithfully to the real world, improve the accuracy of deriving the similarity, and improve robustness when the face state changes.

[0064] The three-dimensional information imaging unit 110 may be a ToF (Time of Flight) camera or an IR (infrared) camera that can obtain distance information. Alternatively, the imaging unit 100 and the three-dimensional information imaging unit 110 may be integrated to obtain brightness information and distance information simultaneously. Alternatively, two imaging units 100 that obtain brightness information, such as a stereo camera, may be used to obtain distance information from the parallax between pixels obtained from the two imaging units.

[0065] 13 is a diagram showing internal processing blocks of the face authentication system 1. Compared to the first embodiment, the internal configuration of the face image registration unit 220 and the operation in the registration flow of the composite image 403 are different. Since the operation of the other blocks is common to the first embodiment, the internal processing of the face image registration unit 220 and the operation in the registration flow of the composite image 403 will be mainly described.

[0066] The following describes the internal processing of the face image registration unit 220. The face image registration unit 220 includes a three-dimensional face landmark detection unit 225, an attachment image processing unit 226, and an authentication data generation unit 223. Compared to the first embodiment, the information to be processed is expanded to three dimensions including distance information in addition to luminance information.

[0067] The bare face image generated by the face area detection unit 210 may be a three-dimensional image having three-dimensional information of brightness information and distance image, or a combination of a two-dimensional image storing brightness information and a two-dimensional image storing distance information, and may be in any image form as long as it has three-dimensional information of brightness information and distance information.

[0068] Fig. 14 is a flowchart for explaining the process of generating and registering a composite image 403 during face registration. In the flowchart of Fig. 14, steps that perform the same operations as in the first embodiment are given the same reference numerals and descriptions thereof will be omitted.

[0069] The bare face image is sent to the three-dimensional face landmark detection unit 225. The three-dimensional face landmark detection unit 225 detects three-dimensional face landmarks based on the distance information acquired by the three-dimensional information imaging unit 110 (S1401). The detection method for the three-dimensional face landmarks in S1401 can use a known method based on geometric pattern matching, a detector based on a NN model, or the like, and the detection method is not limited to this as long as the face landmarks can be acquired three-dimensionally.

[0070] Information on the detected three-dimensional face landmarks is sent to the attachment image processing unit 226 together with the bare face image generated by the face area detection unit 210. The attachment image processing unit 226 acquires the attachment image 402 from the data storage unit 400, processes the attachment three-dimensionally based on the three-dimensional face landmarks (S1402), synthesizes it three-dimensionally so that it fits the face image (S1403), and sends the generated attachment-synthesized face image to the authentication data generation unit 223.

[0071] In S1402, the attachment image 402 obtained from the data storage unit 400 may be in either a three-dimensional image or a two-dimensional image, as long as the attachment can be three-dimensionally processed and synthesized into a three-dimensional face image. The form of the bare face image 401 and the synthesized image 403 recorded in the data storage unit 400 may be a feature vector generated by a feature extractor, a three-dimensional image, or a three-dimensional image converted into a two-dimensional image, or a combination of these. When generating a feature vector, the image form input to the feature extractor may be a three-dimensional image or a two-dimensional image, or a combination of these may be input to the feature extractor to generate an integrated feature vector.

[0072] Fig. 15 is a configuration diagram of the face authentication unit 300 in the present embodiment 3. Compared with Fig. 9 of the embodiment 2, the internal processing of the face image matching unit 320 is different, but the operations of the other blocks are common to the embodiment 2. In the present embodiment 3, the face landmark detection unit 324 and the attachment image processing unit 325 in Fig. 9 are replaced with a 3D face landmark detection unit 326 and an attachment image processing unit 327, respectively.

[0073] Fig. 16 is a flowchart showing the operation during face authentication in embodiment 3. In the flowchart of Fig. 16, steps performing the same operations as in embodiment 2 are given the same reference numerals and descriptions thereof will be omitted.

[0074] If it is determined in S1102 that the wearer is wearing an accessory, the bare face image 401 and the wearer image 402 (S1103) acquired by the wearer image acquisition unit 323 are sent to the three-dimensional face landmark detection unit 326, which detects the face landmarks three-dimensionally (S1401). The detected three-dimensional face landmark information, the bare face image 401, and the wearer image 402 are sent to the wearer image processing unit 327. The wearer image 402 is three-dimensionally processed based on the three-dimensional face landmark information (S1402), and three-dimensionally synthesized to fit the bare face image 401 (S1403).

[0075] <Embodiment 3: Summary> In the face authentication system 1 according to the third embodiment, compared to the first and second embodiments in which face landmarks are detected two-dimensionally and an accessory is synthesized, the face authentication system 1 according to the third embodiment processes and synthesizes an accessory image three-dimensionally based on a three-dimensional face landmark, so that even if the face direction changes, the curved surface of the accessory can be faithfully reproduced, and the accessory image 402 can be synthesized with a bare face image 401 with higher accuracy. This makes it possible to improve the accuracy of deriving the similarity by faithfully reproducing the appearance when the accessory is worn, and to improve the robustness of the similarity depending on the face direction. In addition, since distance information is also included in the matching data during face matching, the authentication accuracy can be improved not only when the accessory is worn, but also when the bare face image is used alone.

[0076] <Fourth embodiment> In the fourth embodiment of the present invention, a configuration for improving the matching efficiency in the first embodiment and speeding up the matching process will be described. In the first embodiment, as the number of variations in wearable items increases, the number of composite images 403 registered in the data storage unit 400 also increases proportionally (particularly in 1:N authentication, the amount of registered face data increases by the number of people to be registered), so if matching is to be performed sequentially for all registered face data stored in the data storage unit 400, the time required for matching increases. In the fourth embodiment, an attempt is made to speed this up.

[0077] FIG. 17 is a configuration diagram of the face authentication unit 300 in the fourth embodiment. The difference from the first embodiment is that a matching category detection unit 328 and a matching data acquisition unit 329 are added to the face image matching unit 320. The advantage of adding the matching category detection unit 328 is that, for example, the presence or absence of an attachment (whether the face is bare or an attachment is attached), the type of attachment (mask, goggles, glasses, hat, etc.), the attributes of the person (age, sex, etc.), the direction of the face (a combination of one or more of yaw, roll, and pitch), etc. can be detected in advance, and the category of registered face data used for matching can be narrowed down to one corresponding to the detected attributes. This improves matching efficiency and speeds up matching time compared to the case where all registered images are authenticated sequentially.

[0078] Fig. 18 is a flowchart for explaining the operation at the time of authentication in this embodiment 4. In the flowchart of Fig. 18, steps performing the same operations as those in embodiment 1 are given the same reference numerals and the explanation thereof will be omitted.

[0079] The face image generated by the face area detection unit 310 is sent to the matching category detection unit 328. The matching category detection unit 328 detects the category of the face image (S1801). Examples of the matching category include the presence or absence of an attachment, the type of attachment, the attributes of the person, and the face direction, or a combination thereof. As a category detection method, a known geometric shape pattern matching, a detector based on a NN model, or a combination thereof can be used. A configuration in which two or more NN model detectors are combined may be used. For example, a combination of a NN model detector that distinguishes the type of attachment and a dedicated NN model detector that determines the gender classification is used. Information on the detected category is sent to the matching data acquisition unit 329. The matching data acquisition unit 329 acquires a registered image (natural face image 401 or synthetic image 403) suitable for the detected category from the data storage unit 400 (S1802) and sends it to the authentication data generation unit 321.

[0080] <Embodiment 4: Summary> The face authentication system 1 according to the fourth embodiment acquires only the natural face image 401 or the composite image 403 corresponding to the type of the accessory or the attribute of the person to be authenticated from the data storage unit 400, and performs face authentication using this. This makes it possible to narrow down the registered face data used during matching and reduce the number of matching operations, thereby improving the matching efficiency and speeding up the matching.

[0081] <Fifth Preferred Embodiment> In the fifth embodiment of the present invention, a configuration is described in which the robustness against the face direction is improved by changing the face direction in order to perform face authentication more robust against the face direction. In applications such as simultaneous face authentication of multiple users using signage and face authentication from surveillance camera images, the face direction at the time of authentication is not limited to the front face, and the variation of the face direction angle is large. Therefore, for example, there are cases where authentication is required when the face is facing diagonally. In the fifth embodiment, it is intended to further improve the robustness of authentication in such cases.

[0082] 19 is a configuration diagram of the face registration unit 200 in the fifth embodiment. The difference from the first embodiment is that a face direction change unit 227 is added to the face image registration unit 220. The advantage of adding the face direction change unit 227 is that a large number of natural face images 401 and composite images 403 with various face directions can be stored in advance in the data storage unit 400, and highly accurate authentication can be realized regardless of the face direction at the time of authentication. Explanation of the blocks common to the first embodiment will be omitted, and the face direction change unit 227, which operates differently, will be explained below.

[0083] The unmasked face image generated by the face area detection unit 210 is first sent to the face direction changing unit 227. The face direction changing unit 227 changes the face direction angle of the input face image. The face image whose face direction has been changed is sent as is to the authentication data generation unit 223 when generating an unmasked face image 401, and is sent to the face landmark detection unit 221 when generating a composite image 403.

[0084] The face direction changing unit 227 can use face direction deformation by a known affine transformation or a non-rigid deformation method based on the FFD method. The FFD non-rigid deformation method displaces control points on a face image and deforms the face image by moving the control points. When face landmarks are used as control points, a face landmark detection unit 221 may be disposed before the face direction changing unit 227, and face landmark detection may be performed first, and then the face direction may be changed. In addition, the face direction changing unit 227 may generate face images in a plurality of directions, and then generate a face image in an intermediate direction from the plurality of face images with different face directions by complementing them. For example, a face image in an intermediate diagonal direction may be generated from a face image in front and a face image in profile.

[0085] FIG. 20 is a diagram showing the effect of the fifth embodiment. Here, as an example, an example is shown in which the face direction is changed in two directions by the face direction changing unit 227, and then a mask image is synthesized as the attachment image 402. Based on the bare face image 20-1 generated by the face area detection unit 210, a bare face image 20-2 tilted at an angle in the roll direction and a bare face image 20-3 tilted at an angle in the yaw direction are generated by changing the face direction. Face landmark detection is performed for each of the bare face images 20-1, 20-2, and 20-3, and the attachment image is processed and synthesized to fit the face direction, thereby generating attachment-synthesized face images 20-4, 20-5, and 20-6, respectively. By recording 20-1 to 20-6 in the data storage unit 400, face authentication corresponding to three face direction angles including before and after the face direction change is possible at the time of authentication. In the example of FIG. 20, the face direction is changed in two directions, but the direction in which the face direction is changed is arbitrary, and the number of directions is not limited.

[0086] The fifth embodiment may be combined with the fourth embodiment to speed up matching during authentication. In detecting a matching category during face authentication, the face direction is detected and registered face data having the same face direction as the detected face direction is obtained from the data storage unit 400, thereby improving matching efficiency.

[0087] FIG. 21 is a configuration diagram of the face authentication unit 300 in the fifth embodiment. The difference from the second embodiment is that a face direction change unit 331 is added to the face image matching unit 320, and the accessory shape pattern detection unit 322 is replaced with the accessory shape pattern detection unit 330. Compared to the second embodiment, the advantage of adding the face direction change unit 331 is that the face direction of the face image registered in the data storage unit 400 can be changed according to the face direction at the time of authentication. This makes it possible to generate a composite image having the same face direction angle as the user's face direction angle at the time of authentication, thereby improving the accuracy of deriving the similarity. Below, the description of the blocks common to the second embodiment will be omitted, and blocks whose operations are different will be described.

[0088] From the face image generated by the face area detection unit 310, the accessory shape pattern detection unit 330 detects the pattern of the accessory worn at the time of authentication and the face direction. Based on the detected face direction information, the face direction change unit 331 changes the acquired bare face image 401. The bare face image 401 with the face direction changed to match the face direction angle at the time of authentication is sent to the face landmark detection unit 324 to process and combine the accessory. If the user was not wearing an accessory at the time of authentication, the bare face image 401 with the face direction changed is sent to the authentication data generation unit 321 as is.

[0089] The accessory shape pattern detection unit 330 detects the face direction in addition to detecting the accessory shape pattern described in S1101 of Fig. 11. The face direction detection can be performed using a method of detecting known face landmarks and estimating the face direction angle by comparing them with the face landmarks when facing forward at a predetermined point, or a detector based on a NN model, etc.

[0090] <Embodiment 5: Summary> In the face authentication system 1 according to the fifth embodiment, in both the first and second embodiments, the face orientation change unit 331 can flexibly change the face orientation angles of the natural face image 401 and the composite image 403 used during authentication, and store the images in advance in the data storage unit 400. This enables highly accurate face authentication even if the user's face is facing in various directions during authentication, and realizes a face authentication system that is more robust against face orientation angles.

[0091] <Modifications of the present invention> The present invention is not limited to the above-described embodiment, and includes various modified examples. For example, the above-described embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to those having all of the configurations described. In addition, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. In addition, it is possible to add, delete, or replace a part of the configuration of each embodiment with another configuration.

[0092] In the above embodiment, an example has been described in which the face registration unit 200 and the face authentication unit 300 are connected to one imaging unit 100, but a configuration in which an imaging unit 100 is provided for each of the face registration unit 200 and the face authentication unit 300 may also be used.

[0093] In the above embodiments, the face recognition system 1 is illustrated as an integrated unit, but the imaging unit 100, face registration unit 200, face recognition unit 300 and data storage unit 400 may each be located in a different spatial location and each may wirelessly communicate data with each other from a data base station 2.

[0094] In the above embodiments, the authentication data generating unit may extract features of a composite image using different feature extractors for different types of accessories (including the presence or absence of accessories), personal attributes, face orientation, or combinations thereof. For example, in the first embodiment, if the person to be authenticated is not wearing accessories, a feature extractor for bare face images is used, and if the person to be authenticated is wearing a mask, a feature extractor for masks is used. Alternatively, in the fourth embodiment, if the attribute of the person to be authenticated is "male", a feature extractor for men is used. Alternatively, if the person to be authenticated is "mask, male", a feature extractor for masks and men is used. Similarly, when registering a face image, different feature extractors may be used for different types of accessories and personal attributes.

[0095] In the above embodiments, the face registration unit 200 and the face authentication unit 300 (and each functional unit arranged as an internal block thereof) can be configured by hardware such as a circuit device that implements these functions, or can be configured by a computing device (e.g., Central Processing Unit: CPU) executing software that implements these functions. [Explanation of symbols]

[0096] 1: Face recognition system 2: Data base station 100: Imaging unit 200: Face registration section 300: Face recognition section 400: Data storage unit 210, 310: face area detection unit 220: Face image registration unit 320: Face image matching unit 401: Real face image 402: Wearing image 403: Composite image 221, 324: Face landmark detection unit 222, 325: Wearable Image Processing Department 223, 321: Authentication data generation unit 224: Face image recording unit 322: Wearing object shape pattern detection unit 323: Wearable image acquisition unit 110: 3D information imaging unit 225, 326: 3D face landmark detection unit 226, 327: Wearable Image Processing Department 328: Matching category detection unit 329: Matching data acquisition unit 227, 331: Face direction change section

Claims

1. A facial recognition system for identifying an individual using a facial image, a synthesis unit for generating a synthetic image by synthesizing an accessory image, which is an image of an accessory worn by the individual, with a bare face image of the individual; a face authentication unit that identifies the individual using the composite image; Equipped with The synthesis unit generates the synthetic image by transforming the attachment image so as to conform to a face shape of the individual; The synthesis unit transforms the accessory image so as to match the facial shape of the individual based on landmarks on the bare face image. A face recognition system characterized by:

2. A facial recognition system for identifying an individual using a facial image, a synthesis unit for generating a synthetic image by synthesizing an accessory image, which is an image of an accessory worn by the individual, with a bare face image of the individual; a face authentication unit that identifies the individual using the composite image; Equipped with The synthesis unit generates the synthetic image by transforming the attachment image so as to conform to a face shape of the individual; The face authentication system further includes a face orientation change unit that changes a face orientation of the bare face image, The synthesis unit performs a face direction change on the attachment image in the same manner as the face direction changed by the face direction change unit, The synthesis unit synthesizes the attachment image, the orientation of which has been changed, with the bare face image to generate the synthetic image. A face recognition system characterized by:

3. the synthesis unit detects an angle at which the face image of the individual is rotated with respect to a state in which the individual faces forward, based on the landmark, for at least one of a yaw angle, a roll angle, and a pitch angle; The synthesizing unit rotates the accessory image according to the detected angle, thereby transforming the accessory image to fit the face shape of the individual.

2. The face authentication system according to claim 1 .

4. The synthesis unit transforms the attachment image so that a vertical size of the attachment image coincides with a vertical size of an attachment position of the landmark at which the attachment is to be worn; The synthesis unit transforms the attachment image so that a horizontal size of the attachment image coincides with a horizontal size of an attachment position of the landmark at which the attachment is to be worn; The synthesizing unit transforms the attachment image so that a vertical size of the attachment image and a vertical size of the attachment position match each other and a horizontal size of the attachment image and a horizontal size of the attachment position match each other for each of the divided regions of the attachment image.

4. The face authentication system according to claim 3.

5. The face authentication system further includes an attachment shape pattern detection unit that detects an attachment shape pattern from the face image of the individual, the face authentication unit identifies the individual using the composite image when a shape pattern of the article of wear is detected from the face image; The face authentication unit identifies the individual using the bare face image when the shape pattern of the article of wear is not detected from the face image.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

6. The face authentication system further includes a data storage unit for storing a plurality of types of the attachment images, The face authentication unit acquires from the data storage unit only the attachment image corresponding to the detected attachment shape pattern among the plurality of attachment images, The synthesis unit generates the synthetic image by using only the attachment image acquired from the data storage unit among the plurality of attachment images.

6. The face authentication system according to claim 5.

7. The synthesis unit acquires, as the unmasked face image, a three-dimensional image including luminance information and distance information of the face image of the individual, The synthesis unit specifies feature points in a three-dimensional coordinate system on the bare face image as the landmarks, The synthesis unit transforms the accessory image in the three-dimensional coordinate system based on the landmark in the three-dimensional coordinate system, thereby matching the accessory image with the face shape of the individual.

2. The face authentication system according to claim 1 .

8. The face authentication system further includes a data storage unit that stores the composite image for each type of the attachment, The face authentication system further includes a category detection unit that detects a type of the wearable item, The synthesis unit acquires from the data storage unit only the synthetic image corresponding to the detected type of the attachment, The face authentication unit identifies the individual using the composite image acquired from the data storage unit.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

9. The face authentication system further includes a data storage unit that stores the raw face image and the composite image for each person attribute, The face authentication system further includes a category detection unit that detects attributes of the individual. The synthesis unit acquires from the data storage unit only the bare face image corresponding to the detected attribute of the individual or the synthetic image corresponding to the detected attribute of the individual; The face authentication unit identifies the individual by using the original face image or the composite image acquired from the data storage unit.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

10. The face authentication system further includes a data storage unit for storing the unmasked face image, The data storage unit stores at least the unmasked face image in which the individual faces forward and the unmasked face image in which the individual faces at an angle.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

11. The wearable article is an object that covers at least a part of the face of the individual when worn by the individual.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

12. The wearable article is at least one of a mask, glasses, goggles, and a hat.

12. The face authentication system according to claim 11.

13. The face authentication system further includes an authentication data generation unit that extracts a feature amount of the composite image as authentication data used to identify the individual, The face authentication unit identifies the individual by comparing a feature amount of the face image of the individual with the feature amount extracted from the composite image.

3. The face authentication system according to claim 1, wherein the face authentication system is a face authentication system.

14. The authentication data generating unit extracts features of the composite image using a feature extractor corresponding to the type of the attachment, a feature extractor corresponding to the attribute of the individual, or a feature extractor corresponding to a face direction.

14. The face authentication system according to claim 13.

Images