JP4611575B2 - Face image collation device for personal authentication - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a face image collation device for personal authentication, and more particularly to a face image collation device for personal authentication that generates a plurality of reference images.
[0002]
[Prior art]
In recent years, there are so-called biometrics methods that identify individuals based on their physical characteristics when managing entry and exit of users in facilities that require a high level of security or when restricting access to confidential data. It's being used. Different physical features such as fingerprints, eyeball irises and faces are used for identification.
[0003]
In a face image collation apparatus using a human face image, a face image photographed by a video camera or the like is compared with a face image registered in advance to authenticate whether or not the person is the person.
[0004]
When the face image collating device is used in the locking system, if the collator is determined to be the person himself / herself, an electric lock release signal is output to permit entry / exit of the facility or the like. On the other hand, if it is determined that the person is not the person, the person who performed the collation is determined to be an impersonator, and an alarm signal is output to the manager room, the security room, or the like. The advantage of the face image matching device is that it is easy to identify a suspicious person by looking at the history of the video camera in the case of unauthorized use.
[0005]
On the other hand, the face image matching device has the following problems. Face images taken with a video camera at the time of authentication vary depending on the collator's posture and ambient lighting conditions that change from time to time.For this reason, for example, even though the collator is the person himself / herself, There has been a problem that the image at the time of collation is significantly different from the image at the time of registration, and the collation device rejects the person. In general, two kinds of error rates are used as a scale for evaluating the performance of the personal verification device: the person rejection rate or the impersonator acceptance rate. The principal rejection rate is the probability that the person will be determined to be an impersonator regardless of the registered person. The impersonator acceptance rate is a probability that a person to be authenticated is determined to be a registered person even though the person is not registered in the verification device. These error rates are generally in a trade-off relationship. When one error rate is lowered, the other error rate is increased. These are desirably reduced as much as possible.
[0006]
In order to reduce the error rate, a technique found in the conventional H4-256185 publication is disclosed. First, a 3D model of the head is constructed from 3D image data of a registered person obtained using a surface measuring device. Next, by changing the posture of the three-dimensional model or changing the parameters representing the light source position and the light source attribute, a plurality of two-dimensional images of the model viewed from the front are generated. A plurality of generated two-dimensional images are registered as reference images, and personal authentication is performed in comparison with images input at the time of authentication. In the above prior art, since a plurality of reference images are prepared, the error rate can be lowered to some extent.
[0007]
[Problems to be solved by the invention]
However, the surface measuring device needs to use a special measuring device other than a normal camera, and its handling is not always easy. In addition to the posture of the collator with respect to the video camera at the time of collation and lighting conditions, there has been a problem that the error rate of the collation apparatus becomes high due to the influence of, for example, a change in hairstyle or a change in makeup.
[0008]
The present invention has been made in view of the above problems, and a first object of the present invention is to provide a face image collation apparatus that allows easy personal identification.
[0009]
A second object of the present invention is to provide a face image matching device with a low error rate.
[0010]
[Means for Solving the Problems]
  A face image matching device according to an aspect of the present invention is a face image matching device used for authentication of a person registered in advance, and is extracted from a target face image that is a two-dimensional face image input as an authentication target. In a face image matching apparatus that performs personal authentication by comparing target data with reference data extracted from a reference face image that is a two-dimensional face image used for matching of the target face image, the person used for identity registration Personal registration image input means for inputting a two-dimensional face image, a standard frame model representing a three-dimensional shape of a standard face, and a plurality of rendering conditions in consideration of factors of face variation that can be assumed at the time of authentication. Storage means for storing in advance, personal model generating means for generating a 3D face model corresponding to the face of the person by combining the 2D face image of the person and the standard frame model, and the third order Reference face image generation means for generating a plurality of reference face images according to the plurality of rendering conditions stored in the storage means from a face model, and each of the plurality of reference face images generated by the reference face image generation means Collating means for determining whether or not the target face image is the face image of the person, using a result of collating each of the extracted reference data and the target data extracted from the target face image; , A face image collating device for personal authentication.
In the face image matching device according to one aspect of the present invention, the plurality of rendering conditions may include at least one of an attitude, a light source position, and a light source attribute of the three-dimensional face model.
In the face image collation apparatus according to one aspect of the present invention, the reference face image generation unit includes an individual model deformation unit that deforms the three-dimensional face model based on each face change pattern, and the deformed three-dimensional face A plurality of reference face images can be generated based on a model, and the change pattern includes color information corresponding to the type of makeup and control information related to its distribution shape, and control related to changes in facial expression muscles corresponding to emotions Information.
In the face image matching device according to one aspect of the present invention, a skin color calculation unit that calculates a skin color component of the face from the authentication face image, and a skin color that is not a skin color in the authentication face image based on the skin color component And a removing unit that identifies the region and excludes the region from the verification target range.
In the face image matching device of one aspect of the present invention, the skin color calculation unit extracts a region where a color component of a color system corresponding to a skin color is within a predetermined range from the authentication face image, and averages luminance in this region The value may be calculated as the skin color component.
The description from the next sentence to the paragraph [0019] of this paragraph corresponds to the description of [Claims] at the beginning of the application.
  In order to achieve the above object, a face image matching device for personal authentication according to the present invention is a face image matching device used for authentication of a registered person, and is a two-dimensional face input as an authentication target. Face image for performing personal authentication by comparing target data extracted from a target face image that is an image and reference data extracted from a reference face image that is a two-dimensional face image used for matching the target face image In the collation device, a personal registration image input means for inputting a two-dimensional face image of the principal used for personal registration, a standard model storage means for storing a standard frame model representing a three-dimensional shape of a standard face, Personal model generation means for generating a 3D face model corresponding to the face of the person by synthesizing the 2D face image of the person and the standard frame model, and a plurality of renderins from the 3D face model Characterized in that it and a reference face image producing means for producing a plurality of the reference face image according to the conditions.
[0011]
According to the above configuration, a plurality of reference face images can be generated according to a plurality of rendering conditions, so that the error rate can be reduced even when the posture of the person of the target face image and the illumination conditions change. Here, the rendering condition is a condition used for generating a reference face image from the three-dimensional face model, and relates to the posture, light source position, light source attribute, and the like of the three-dimensional face model.
[0012]
The 3D face model corresponding to the person's face is generated by combining the 2D face image of the person and the standard frame model, but the 2D face image of the person is obtained by using a special measuring device such as a surface measuring device. There is no need to use it, and identity registration can be performed easily. The standard frame model is a three-dimensional model representing a three-dimensional shape of a standard human face, and a wire frame model, a surface model, or the like is used. The standard frame model may be configured to have a standard face image on the surface.
[0013]
In another aspect of the present invention, there is provided personal model deforming means for deforming the three-dimensional face model based on each face change pattern, and a plurality of the reference face images are obtained based on the deformed three-dimensional face model. It can be generated. The change pattern relates to a person's posture in the target face image and a variation factor other than the illumination condition, and relates to a facial expression, makeup (eyeline, lipstick, etc.), for example. Since a plurality of reference face images can be generated based on the change pattern, it is possible to reduce the error rate related to facial expression of the target face image, makeup (eyeline, lipstick, eyebrows, etc.) and the like. Preferably, a change pattern setting unit for setting the change pattern is provided.
[0014]
In a further aspect, the face image matching device for personal authentication according to the present invention includes a presentation unit that presents a plurality of reference face images generated by the reference face image generation unit, and a plurality of the reference face images that are presented. And a reject image designating unit for designating a face image not used for collation, and rejecting the reference face image designated by the reject image designating unit. According to the above configuration, it is possible to leave only the reference face image that is considered necessary at the time of collation from the presented reference face images, so that it is possible to perform authentication efficiently without using extra reference data at the time of registration. it can.
[0015]
Preferably, the standard model storage means stores a plurality of standard frame models, and the personal model generation means includes attribute information input means for inputting the user's attribute information, and the attribute information based on the attribute information. And a standard model selection means for selecting one standard frame model from a plurality of standard frame models. The attribute information is, for example, the contour shape, age, gender, etc. of the person's face. According to the above configuration, a suitable standard frame model can be selected according to the attribute information.
[0016]
Further, the personal model generating means extracts an image of a predetermined part of the face from the two-dimensional face image of the person, and obtains the three-dimensional shape of the standard frame model based on the image of the predetermined part. And means for deforming.
[0017]
According to the above configuration, since the standard frame model can be deformed using the image area of the part necessary for collation, the processing speed is improved.
[0018]
In order to achieve the second object, a face image collation device for personal authentication according to the present invention is a face image collation device for performing authentication of a person registered in advance, and the input authentication face. In a face image collation device for performing personal authentication by collating an image with a registered reference face image, a skin color calculation means for calculating a skin color component of the face from the authentication face image, and a skin color component And a removing unit that identifies a region that is not a skin color in the authentication face image and excludes the region from the verification target range.
[0019]
According to the above-described configuration, the skin color component of the face of the authenticator calculated by the skin color calculation means is a non-skin color portion of the face image, which causes an error rate at the time of matching, for example, glasses, hair, wrinkles Etc. can be excluded from the collation target range.
[0020]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention (hereinafter referred to as embodiments) will be described with reference to the drawings.
[0021]
The face image collation device for personal authentication according to this embodiment is used in an access control system for a room in which an electric lock is installed at an entrance.
[0022]
[First embodiment]
FIG. 1 is a block diagram illustrating an overall configuration of an access control system 12 including a face image matching device 10 according to the first embodiment.
[0023]
Reference numeral 14 denotes an input means which detects the approach of the user by pressing a button (not shown) provided in the access control system 12 or an ultrasonic sensor (not shown) by the user. A video signal of a human face image taken by the camera 16 capable of output is received and output to the control means 18.
[0024]
Reference numeral 20 denotes selection input means, which receives a signal in which an image presented by the user is selected by the numeric keypad 22 or the like, or input of individual attribute information (gender, age, etc.) is received and output to the control means 18. Here, a touch panel or the like may be used instead of the numeric keypad 22.
[0025]
Reference numeral 24 denotes a storage means constituted by a memory. As will be described later, a rendering condition used when generating a plurality of reference face images from a three-dimensional face model, a change pattern for deforming the three-dimensional face model, and a standard A typical three-dimensional shape of a face, a plurality of standard models having face images on the surface thereof, and a plurality of reference face images generated using these models are stored. In addition, the feature amount (reference data) of the person's face extracted from the reference face image used for collation is stored. Here, the standard model is a wire obtained from face images of a plurality of persons normalized to a predetermined size using the inclination of a straight line connecting both eyes and the distance between both eyes with a person facing the camera. It is a three-dimensional face model created by averaging the frame model in consideration of the composition ratio of age and gender, and is a three-dimensional model having a face image averaged on the surface.
[0026]
The control unit 18 includes a registration unit 26, a verification unit 28, an electric lock control unit 30, and a feature amount extraction unit 32.
[0027]
The registration unit 26 is used at the time of registration, and generates a plurality of reference face images for the inputted face image based on the standard model, rendering conditions, and change patterns stored in advance and stores them in the storage unit 24. Remember.
[0028]
The matching unit 28 is used when matching a person, compares each stored reference face image with the target face image to be authenticated input at the time of matching, and calculates a plurality of similarities for each part of the face. To do. The calculated similarity is comprehensively determined, and when the face image is the registrant's own face, the electric lock 36 is unlocked via the electric lock control means 30 and the output means 34. If it is not the person's face image, the electric lock is not unlocked, and for example, the input face image is stored in the storage unit 24 together with the date and time information of the photographed image.
[0029]
The feature amount extraction unit 32 extracts a feature amount (target data) that is extracted from the target face image to be authenticated input at the time of matching and is used for matching. The feature amount (reference data) extracted from the reference face image is stored in advance in the storage unit 24 and compared with the feature amount extracted from the target face image input at the time of collation. As the feature amount, for example, luminance information, edge strength, edge gradient, and the like of each part of the face such as eyes and mouth included in the face image are used. It is also possible to use a feature amount extracted from the entire face image instead of every predetermined part of the face.
[0030]
Reference numeral 34 denotes output means for displaying a plurality of reference face images generated by the registration means 26 on the monitor 38 and controlling the electric lock 36 based on the result of the electric lock control means 30.
[0031]
The operation in the personal registration mode of the face image matching device 10 configured as described above will be described with reference to FIG. FIG. 2 shows a detailed configuration of the registration unit 26.
[0032]
Here, the personal registration mode is a mode in which a personal face image input at the time of personal registration and a face image that considers the variation factors of the personal face image are registered in the collation apparatus 10 as a reference face image. When using this mode, the administrator having the authority of the collation apparatus 10 shifts from the mode in which the normal collation operation is performed to the mode.
[0033]
First, a two-dimensional face image of the registrant is photographed with the camera 16 or the like. Next, the captured image is displayed on a monitor 38 (see FIG. 1) or the like, and the registrant or the administrator visually determines that the registrant faces the camera and the face is not tilted. Then, the image in that state is input to the registration unit 26 via the input unit 14. At that time, for example, a horizontal line as a reference for the position of both eyes of the face may be displayed on the monitor 38 so that a face image with a tilted face is not input. Next, the input face image is normalized to a predetermined size using the slope of the straight line connecting both eyes and the distance between both eyes. Hereinafter, this normalized face image is referred to as a personal registration image. A plurality of reference face images are generated in consideration of the face variation factors that can be assumed at the time of authentication for the personally registered image.
[0034]
First, the part extraction unit 42 extracts the head region and each part region of the face from the face registration image and the face image of the standard model.
[0035]
For example, when a head region is extracted from a face image, a difference image between the face image and a background image in which a person photographed in advance is not captured is created and binarized using a predetermined threshold value. By calculating vertical and horizontal histograms from the obtained two-dimensional image, a binarized head region image of the registrant can be obtained. By calculating the logical product of this image and the face image, the head region can be extracted from the face image.
[0036]
Next, an eye area is extracted. Since the eyes have a feature that the horizontal edge is strong in the face image, the edge strength and the relative positional relationship in the head region (for example, above the center in the horizontal direction of the head region and the head) The eye area is extracted as a square area having a constant area by combining a straight line in the center of the partial area, which is substantially symmetrical with respect to the straight line in the vertical direction and having a strong horizontal edge.
[0037]
Next, the nose and lip regions are extracted. With respect to these, taking into account the relative position from the eye area (for example, a predetermined distance away from the lower center of the left and right eye areas), the area is extracted as a square area having a constant area.
[0038]
In addition, each said part area | region is shifted and matched using the template of the average part area | region previously memorize | stored in a head area | region, and the area | region with the highest similarity shall be made into the area | region where an object part exists. Is also possible.
[0039]
Next, the three-dimensional face model generation unit 44 transforms the corresponding standard part of the standard model stored in advance based on the user's head region, the shape of each region, and the relative positional relationship between them. Then, a 3D face model of the registrant is generated.
[0040]
Specifically, nonlinear transformation is performed so that the size of the head region of the face image that the standard model has, the size of the edges extracted from each region, and the positional relationship thereof match those of the registered user image. After the application, each part region of the standard model is replaced with a corresponding part region of the personal registration image. As a result, a three-dimensional face model of the registered person image is obtained.
[0041]
Next, the reference image generation unit 46 considers the generated three-dimensional face model, previously stored rendering conditions, and deformation patterns, and a reference face image that is a two-dimensional image viewed from the same viewpoint as the standard model. Is generated.
[0042]
Here, what is considered as the rendering condition is a variation factor such as the posture of the registrant himself, the light source position, the light source attribute, and the like. For example, when considering variation due to posture, a total of 19 reference images are generated using conditions in which the head is rotated to ± 15 ° in 5 ° increments in the pitch, yaw, and roll directions of the three-dimensional face model. .
[0043]
Further, the variation factors of the face considered as the deformation pattern are due to makeup and facial expressions. In the case of makeup, the specific content of the deformation pattern is control information relating to the type of makeup (color information and its distribution shape, etc.) for eyelines, lipsticks, eyebrows, etc. for each part of the eyes, lips, and eyebrows. In the case of a change in facial expression, for example, control information relating to a change in facial expression muscle corresponding to emotions in a wire frame model. Based on these control information, the reference image generation means 46 deforms the three-dimensional face model to generate a plurality of reference face images.
[0044]
The reference face image presenting means 48 presents a plurality of reference face images generated by the reference face image generating means 46 to the registrant using the monitor 38 or the like. A registrant selects a reference face image to be used for collation with a numeric keypad or the like in consideration of his / her preference (for example, the color and shape of the lipstick, which is easy to hold the camera slightly to the left). This information is input to the reference face image designating unit 50, and is stored in the storage unit 24 by the reference face image registering unit 52 as a reference face image together with the identity registration image. On the other hand, the reference face image that was not selected is rejected as not being used at the time of matching. The reference face image designating unit 50 may select a reference face image to be rejected.
[0045]
In addition, a change pattern such as makeup (lipstick, eyeline, eyebrow, etc.) and facial expression stored in the storage means 24, or a reference face image to which these are given is presented on the monitor 38, and the user is provided with his / her own taste. It is also possible to select a match with the numeric keypad 22 or the like.
[0046]
In addition, by causing the registrant to input attribute information such as sex and age by using the numeric keypad 22 or the like, for example, when the registrant is a woman, an appropriate deformation pattern such as changing a makeup pattern to be presented according to the age and fashion. By selecting, it is also possible to efficiently generate a reference face image.
[0047]
Next, a weighting factor for each individual used for similarity calculation at the time of collation is obtained as an amount representing how much each feature amount represents the individuality of the person.
[0048]
Specifically, the variance is obtained from the feature amount obtained for each part of the plurality of reference face images, and a large weight is given to the feature amount having a small value. In addition to this, the feature values obtained from the reference face image of the registrant in a general pattern classifier such as the fisher method or Support Vector Machine using a separately prepared impersonator database and the number of non-registrants stored in the database It is also possible to input a feature amount obtained from the three face images and give a weight that maximizes the identification rate of these face images.
[0049]
The extracted feature quantity is stored in the storage unit 24 together with the weighting coefficient obtained for each registrant as reference data used at the time of collation.
[0050]
When the registration process ends, the administrator switches the operation mode to the verification mode.
[0051]
Next, the process at the time of collation is demonstrated.
[0052]
First, a target face image to be authenticated is input, and a plurality of part regions are extracted by the part extracting means 42 and their feature amounts are calculated.
[0053]
By calculating the similarity between the calculated feature quantity and the feature quantity obtained from each reference face image, and multiplying them by the individual weight coefficient obtained at the time of registration, The similarity is calculated.
[0054]
Specifically, each part region image is extracted from 19 reference face images and target face images having different postures generated at the time of registration, and a luminance difference or correlation between pixels included in the image is calculated for each part. A similarity (hereinafter referred to as a part similarity) is obtained.
[0055]
The sum of all the part similarities multiplied by the individual weighting coefficient obtained at the time of registration is obtained, and this is used as the similarity in the entire reference face image (hereinafter referred to as face image similarity). Calculation is performed for all reference face images, and then comprehensive determination is performed using the obtained plurality of face image similarities.
[0056]
As a result of the comprehensive determination, if it is determined that the person is the person, for example, a release signal is sent to the electric lock of the door. If the person is not identified, the electric lock is not released.
[0057]
It is also possible to update the reference face image and the personal weighting coefficient using the face image determined as the person at the time of collation.
[0058]
For example, by examining which of the plurality of reference face images the target face image received as the person is most similar to, and repeating the registration process again with the most similar reference face image as the person registration image Update registration of reference face image. As a result, since the reference face image is replaced with a face image that is actually authenticated (actually changed face image), it is possible to obtain a reference face image that more reflects the change in the face image realized at the time of authentication. .
[0059]
Also, for example, the frequency determined to be the most similar for each reference face image is stored, and the weighting coefficient is recalculated with emphasis on the features obtained from the reference face image having a high frequency, thereby updating the personal weight coefficient. It is also possible to do this.
[0060]
In the present embodiment, a single standard model obtained by averaging a plurality of three-dimensional models and face images is used. However, the standard model is not limited to this, for example, gender, age, or It is also possible to use a plurality of models averaged for each typical three-dimensional contour shape of the face. In this case, at the time of registration, user attribute information (for example, gender, age, nose height, etc.) is input using the numeric keypad 22 and the like, and one of a plurality of standard models based on the input attribute information. And a reference face image may be generated based on the standard model.
[0061]
In the present embodiment, the storage unit 24 stores a plurality of reference face images. However, only the personal registration image and information specifying the reference face image used for collation are recorded and authenticated. In addition, a reference face image designated by the information may be generated. Further, only reference data may be stored. In this way, the capacity of the memory can be used efficiently.
[0062]
[Second Embodiment]
Next, an example will be described in which an image region that is a variation factor such as wrinkles, glasses, and hairstyle included in the authentication target face image is excluded based on the skin color information of the face image.
[0063]
FIG. 3 is a block diagram illustrating the overall configuration of the access control system 12 including the face image matching device 10 according to the second embodiment. In addition, the same code | symbol is attached | subjected about the part which is common in FIG. 1, and description is abbreviate | omitted.
[0064]
Reference numeral 14 denotes an input means, which detects the approach of the user by pressing a button (not shown) provided in the access control system 12 or an ultrasonic sensor (not shown) by the user, and is analog or digital. A video signal of the person's face image taken by the color camera 54 capable of output is received and output to the control means 18.
[0065]
Reference numeral 24 denotes a storage unit including a memory, which stores a reference face image and a feature amount (reference data) used for matching extracted from the face image.
[0066]
Reference numeral 18 denotes a control means, which comprises a registration means 26, a collation means 28, an electric lock control means 30, a feature amount extraction means 32, and a fluctuation removal means 56.
[0067]
The registration unit 26 stores the face image input at the time of registration in the storage unit 24 as a personally registered face image. The fluctuation removing unit 56 removes fluctuation factors such as wrinkles, glasses, and hairstyles by removing image areas other than the skin color using the skin color information calculated from the face image.
[0068]
Next, the fluctuation removing unit 56 will be described in detail with reference to FIG.
[0069]
The part extracting means 58 detects the approach of the user by pressing the button by the user or by an ultrasonic sensor installed separately, and the like from the color target face image to be authenticated taken by the color camera. A plurality of image regions including a part are extracted. Since the specific extraction method is the same as that of the part extraction means 42 in the first embodiment, the description thereof is omitted here.
[0070]
At this stage, the influence of spectacles, wrinkles, etc. can be considered, so there is a possibility that the predetermined part of the face cannot be extracted accurately. However, since the purpose of this process is to obtain a region for calculating the color component of the reference skin color of the face image by the reference skin color calculation means 62, for example, a template for each part region is created in a large size, It is preferable that the approximate position can be extracted.
[0071]
The color separation unit 60 uses a color filter corresponding to each color system to convert the color target face image to be authenticated obtained from the input unit 14 into a plurality of color systems (YIQ color system, XYZ color system). , HVS color system, etc.) to extract luminance information for each color component. As a result, luminance information, that is, coordinate values in each color space can be obtained for each component of each color system.
[0072]
The reference skin color calculation means 62 calculates the skin color component of the authenticator's face. Because the skin color of the face varies from person to person, such as fair skin, blackish black, and red, the skin color area and non-skin color area should be extracted with high accuracy if the skin color component is fixed. I can't. Therefore, in the face image matching device according to the present embodiment, a skin color component is extracted for each individual, and a skin color region and a non-skin color region are separated on the basis of this color component.
[0073]
Specifically, an area in which a color component (for example, H component (referred to as hue)) of a color system (for example, HVS color system) corresponding to the skin color is within a predetermined range is extracted from the face image. The reference skin color is calculated as the average value of the luminance.
[0074]
The non-skin color area removing unit 64 includes a color component in each color system of a region excluding the part area extracted by the part extracting unit 58 from the input color target face image to be authenticated, and the reference skin color calculating unit. The color component of the reference skin color of the face of the person to be authenticated calculated in 62 is compared, and an image region where the color component is equal to or greater than a predetermined threshold is extracted from the color target face image as a non-skin color region. By replacing this region with the reference skin color or a color component near the region, the variation factor is removed from the face image.
[0075]
As a result, fluctuation factors such as hair, wrinkles, and eyeglass frames can be removed from the target face image, so that the part and feature amount can be extracted with higher accuracy when compared with the reference face image.
[0076]
As described above, in the first and second embodiments, the case where only the person is registered has been described as an example, but the present invention is not limited to this. When a plurality of people are registered, it is possible to perform a brute force comparison with respect to the reference face images of all registrants. In addition, when the number of registered persons increases or when a large amount of deformation patterns are prepared, it takes a long time for the verification process. It is also possible to select and collate with this.
[0077]
Furthermore, the face image input at the time of collation can be saved as a history together with the shooting date and time, and the user can be notified by voice or image that the shot face image has been recorded. As a result, it is possible to expect a deterrent effect against fraud, and high security can be ensured. In addition, even if a misrepresentation by a third party occurs, the stored face image can be used for the criminal investigation.
[0078]
【The invention's effect】
Since it is configured as described above, the face image matching device for personal authentication according to the present invention can easily perform personal registration. Further, it is possible to provide a face image matching device with a low error rate.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an overall configuration of an access control system including a face image matching device according to a first embodiment.
FIG. 2 is a diagram showing a configuration of registration means.
FIG. 3 is a diagram illustrating an overall configuration of an access control system including a face image matching device according to a second embodiment.
FIG. 4 is a diagram illustrating a configuration of a fluctuation removing unit.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Face image collation apparatus, 12 Access control system, 14 Input means, 16 Cameras, 18 Control means, 22 Numeric keypad, 24 Memory | storage means, 26 Registration means, 28 Collation means, 30 Electric lock control means, 32 Feature-value extraction means, 34 Output means, 36 Electric lock, 38 Monitor, 42 Part extraction means, 44 Three-dimensional face model generation means, 46 Reference face image generation means, 48 Reference face image presentation means, 50 Reference face image designation means, 52 Reference face image registration means 54 color camera, 56 variation removing means, 58 part extracting means, 60 color separating means, 62 reference skin color calculating means, 64 non-skin color area removing means.

Claims (5)

A face image matching device used for authentication of a person registered in advance, and used for matching target data extracted from a target face image, which is a two-dimensional face image input as an authentication target, and the target face image A face image collation apparatus that performs identity authentication by collating reference data extracted from a reference face image that is a two-dimensional face image.
A person registration image input means for inputting a two-dimensional face image of the person used for the person registration;
And the standard frame model representing the standard three-dimensional shape of the face, and a plurality of rendering conditions and, you prestores Symbol憶means considering variation factors of the face that can be assumed at the time of authentication,
Personal model generation means for generating a 3D face model corresponding to the face of the person by combining the 2D face image of the person and the standard frame model;
Reference face image generation means for generating a plurality of reference face images from the three-dimensional face model according to the plurality of rendering conditions stored in the storage means ;
Using the result of collating each of the reference data extracted from each of the plurality of reference face images generated by the reference face image generating means and the target data extracted from the target face image, the target face Collating means for determining whether the image is the face image of the person,
A face image matching device for personal authentication, comprising: The face image matching device according to claim 1,
  The face image matching device for personal authentication, wherein the plurality of rendering conditions include at least one of a posture, a light source position, and a light source attribute of the three-dimensional face model. The face image matching device according to claim 1,
The reference face image generation means includes
An individual model deforming means for deforming the three-dimensional face model based on each face change pattern;
A plurality of reference face images can be generated based on the deformed three-dimensional face model;
The change pattern includes color information corresponding to the type of makeup and control information relating to the distribution shape thereof, and control information relating to changes in the facial muscles corresponding to emotions. Verification device. In the face image collation device according to any one of claims 1 to 3, further
Skin color calculating means for calculating a skin color component of the face from the authentication face image;
A removal unit that identifies a region that is not a skin color in the authentication face image based on the skin color component, and excludes the region from the verification target range;
Face image collation device for authentication, characterized in that it comprises a. The face image matching device according to claim 4 ,
The skin color calculation means extracts a region where a color component of a color system corresponding to the skin color is within a predetermined range from the authentication face image, and calculates an average value of luminance in this region as the skin color component. A face image matching device for personal authentication, which is a feature.

Images