KR20090119127A - Method and apparatus for authentication by using face recognition - Google Patents

Abstract

PURPOSE: A certification method and device using face recognition is provided to perform the adult certification by recognizing the face of adults or the face of minors from video data. CONSTITUTION: An image shooting unit(210) obtains the image of the photographed subject. A controller(220) detects facial image or non-facial image from input image data by using face detection delimiter based on face detection algorithm. If the facial image is detected, the controller confirms the face of the adult using the face detection delimiter based on face detection algorithm. The controller performs the certification of the adult.

Description

Method and Apparatus for Face Recognition {Method And Apparatus for Authentication by Using Face Recognition}

The present invention relates to an authentication method and apparatus using face recognition. More specifically, it generates input image data of a user's face image as a subject, extracts a face image from the input image data using a face detection delimiter based on a face detection algorithm, and based on an adult discrimination algorithm. The present invention relates to an authentication method and apparatus using face recognition for performing adult authentication, by checking whether an extracted face image is an adult face or a minor face using an adult discriminator.

In recent years, as computer, electronic, and communication technologies have made rapid progress, interest in technology for recognizing and processing human faces has increased. Accordingly, many application fields such as an access control system using face recognition and a face database search have been applied and developed.

As a conventional technique using face extraction, there is a method of extracting a facial region by analyzing specific elements of a face.

In other words, it is a method of dividing the entire face area into one feature area, and it is a high stability in extracting the face area by decomposing it into positional features and main feature points in consideration of correlations such as eyes, nose, and mouth. However, there is a problem that the algorithm is complicated and takes a lot of processing time.

In addition, a conventional technique for extracting a face includes a method of extracting an entire face region as one feature region.

That is, the method of extracting the entire face region as one feature region is an approach to process the entire face region as an input pattern without extracting individual features, and it is possible to generate complex backgrounds with high accuracy and fast recognition time for simple images. In the case of an image, the performance tends to be lowered.

As described above, since the conventional conventional face recognition method is a method of pre-registering a user's face and detecting and authenticating the registered face, only the registered user can use the corresponding technology and apply the technology to an unspecified number. There was a disadvantage of not being able to.

Meanwhile, in the conventional adult authentication method, the user's real name is authenticated by inputting his or her personal information such as his / her name, telephone number, address, or resident registration number to prove that the user is an adult, or use identification cards such as resident registration card, driver's license, and passport. There was a way to be certified as an adult.

However, the conventional adult authentication method not only has a disadvantage in that the user must possess an identification card, but also has a risk of exposing personal information such as a name, telephone number, address or social security number to be stealed by others. .

In order to solve the above-mentioned problems, the present invention generates input image data of a user's face image as a subject, extracts a face image from the input image data using a face detection identifier based on a face detection algorithm, Check whether the extracted face image is an adult face or a minor face by using an adult discrimination identifier based on an adult discrimination algorithm, and if the result is an adult face, authentication using face recognition to perform adult authentication Its purpose is to provide a method and apparatus.

The present invention to achieve the above object, (a) receiving the input image data; (b) checking whether the input image data is a face image or a non-face image by using a face detection identifier based on a face detection algorithm; (c) If the input image data is the face image as a result of checking in the step (b), whether the face image is an adult face or a minor face using an adult discrimination or discriminator based on an adult discrimination algorithm; Confirming; And (d) when the face image is the adult face, as a result of confirming the step (c), performing an adult authentication.

In addition, according to another object of the present invention, the image capturing unit for obtaining an image photographing the subject; Generates input image data using the image capturing unit, and checks whether the input image data is a face image or a non-face image by using a face detection identifier based on a face detection algorithm. When the data is the face image, an adult discrimination or discriminator based on an adult discrimination algorithm is used to determine whether the face image is an adult face or a minor face. As a result of the check, the face image is the adult face. If, the control unit for controlling to perform the adult authentication; And a storage unit which stores one or more pieces of information among the input image data, the face detection separator, the face image, and the adult discrimination separator.

As described above, according to the present invention, input image data obtained by photographing a user's face image as a subject is generated, a face image is extracted from the input image data using a face detection identifier based on a face detection algorithm, and an adult It is possible to check whether the extracted face image is an adult face or a minor face using an adult discrimination or discriminator based on a discriminant discrimination algorithm, and if it is an adult face, adult authentication can be performed.

Further, according to the present invention, the user does not need to pre-register his or her face for adult authentication using face recognition, and determines whether the user is an adult by simply looking straight toward the image photographing unit of the adult authentication device. Since it can be performed, there is an effect that the convenience of the user is increased.

In addition, according to the present invention, the face image extracted by using the adult discriminator separator after extracting the face image of the user without comparing the user's face and the registered face image for adult authentication using the face is an adult face or Since it is determined whether the face is a minor face, the processing time is shorter than the authentication method using the general face recognition.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to the components of each drawing, it should be noted that the same reference numerals are used as much as possible even if displayed on different drawings. In addition, in describing the present invention, when it is determined that the detailed description of the related well-known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

1 is a block diagram schematically illustrating an authentication system using face recognition according to an embodiment of the present invention.

An authentication system using face recognition according to an embodiment of the present invention includes a subject 110 and an adult authentication device 120.

The subject 110 is an object to be photographed. In the present invention, since the user photographs an image corresponding to a face for adult authentication, the subject 110 will be described in the same concept as the face of the user.

The adult authentication device 120 according to an embodiment of the present invention generates input image data photographing the subject 110, and input image data is a face image or a non-face using a face detection identifier based on a face detection algorithm. Check whether the image is an image, if the input image data is a face image, whether the face image is an adult face or a minor face by using an adult discrimination or discriminator based on an adult discrimination algorithm, As a result of the check, when the face image is an adult face, an adult authentication function is performed.

In addition, the adult authentication device 120 according to the embodiment of the present invention generates multi-step resolution reduction data in which the resolution of the input image data is reduced in stages, and the search image window 410 for each reduction step of the multi-step resolution reduction data. Generates image search data for each part searched at a constant pixel interval, and checks whether the image search data for each part is a face image or a non-face image by using the face detection delimiter. In the case of a face image, a function of checking image search data for each part as a face image is performed.

In addition, the adult authentication device 120 according to an embodiment of the present invention extracts the image pattern from the input image data, and confirms whether the image pattern corresponds to the face image using the face detection delimiter, and as a result, the image When the pattern corresponds to the face image, the input image data is identified as the face image.

More detailed description of the adult authentication device 120 according to an embodiment of the present invention will be described in FIG.

2 is a block diagram schematically showing an adult authentication device according to an embodiment of the present invention.

The adult authentication device 120 according to the embodiment of the present invention includes an image capturing unit 210, a control unit 220, and a storage unit 230.

In the present invention, the adult authentication device 120 is described as including only the image capturing unit 210, the control unit 220 and the storage unit 230, but this is merely an example of the technical idea of the present invention As those skilled in the art to which the present invention pertains, various modifications and variations to the components included in the adult authentication device 120 may be applied without departing from the essential characteristics of the present invention.

The image capturing unit 210 is a means for obtaining image information. The image capturing unit 210 acquires an image of the object 110 and transmits the image to the controller 220.

The control unit 220 according to an embodiment of the present invention is a control means for controlling the overall function of the adult authentication device 120, generates input image data using the image capturing unit 210, based on a face detection algorithm One face detection delimiter is used to determine whether the input image data is a face image or a non-face image, and when the input result is the face image, an adult discriminator based on an adult discrimination algorithm is used. If the face image is an adult face or a minor face using the image, and as a result of the check, if the face image is an adult face, the control performs a function of performing adult authentication.

In addition, the control unit 220 according to an embodiment of the present invention generates multi-step resolution reduction data in which the resolution of the input image data is reduced in stages, and the search image window 410 is fixed for each reduction step of the multi-step resolution reduction data. Generates image search data for each part searched at pixel intervals and checks whether the image search data for each part is a face image or a non-face image by using a face detection delimiter, and as a result, the image search data for each part is a face image. In this case, the control performs a function to check the image search data of each part as a face image.

In addition, the control unit 220 according to an embodiment of the present invention extracts the image pattern from the input image data, and confirms whether the image pattern corresponds to the face image using the face detection delimiter, and as a result, the image pattern is When the face image corresponds to the face image, the control performs a function to check the input image data as the face image.

In this case, the face detection algorithm is preferably used with the Adaboost algorithm, the adult identification algorithm is preferably used with the SVM (Support Vector Machine) algorithm.

Of course, these are merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains are face detection algorithms or adult discrimination algorithms without departing from the essential characteristics of the present invention. It will be applicable to various modifications and variations.

Here, the face detection delimiter is probabilistic reference data for distinguishing a face image or a non-face image, and the adult discrimination discriminator is probabilistic reference data for distinguishing an adult face or a minor face.

The storage unit 230 is a storage means for storing various data necessary for driving the adult authentication device 120. The storage unit 230 stores information of at least one of input image data, a face detection separator, a face image, and an adult identification separator. do.

3 is a flowchart illustrating an authentication method using face recognition according to an embodiment of the present invention.

The adult authentication device 120 generates input image data of photographing the subject 110 using the image capturing unit 210 (S310).

The adult authentication device 120 generates multi-stage reduced data by gradually reducing the resolution of the input image data (S320).

For example, assuming that the resolution (size) of the input image data is 320x240 and the size of the search image window 410 is 16x16, the adult authentication device 120 gradually sets the resolution (size) of the input image data. (Ie 320x240-> 200x150-> 160x120…) to generate multi-level reduced data.

The adult authentication device 120 generates search data for each part searched at a predetermined pixel interval in the search image window 410 for each reduction step of the multi-level reduction data (S330).

For example, the adult authentication device 120 generates image search data for each part searched at predetermined pixel intervals for each reduction step of the multi-level resolution reduction data by the corresponding size of the search image window 410.

The adult authentication device 120 extracts an image pattern of input image data (S340).

The adult authentication device 120 distinguishes whether the image search data for each part is a face image or a non-face image by using a face detection identifier and an image pattern based on a face detection algorithm (S350).

The adult authentication device 120 checks whether the image search data for each part is a face image (S360).

As a result of checking in step S360, when the image search data for each part is a face image, the adult authentication apparatus 120 determines whether the face image is an adult face or a minor face using an adult discrimination or discriminator based on an adult discrimination algorithm. Check (S370).

For example, when an adult identification algorithm is used for adult authentication, an adult discrimination identifier may be obtained using a plurality of face images of 0-18 years old and a plurality of face images of 19 years or older.

That is, the adult authentication device 120 may determine which face image belongs to a plurality of face images between 0 and 18 years old and a plurality of face images over 19 years old using an adult discrimination identifier.

As a result of checking in step S370, if the face image is an adult face, the adult authentication device 120 performs adult authentication (S380).

Adult authentication device 120 is coupled with the vending machine, such as cigarettes, alcoholic beverages to prevent the sale of adult goods to minors.

For example, an image corresponding to a user's face is obtained by using a face detection identifier based on a face detection algorithm from input image data input through the image capturing unit 210 provided in the front of the adult authentication device 120. The detected face image is used to identify an adult face by using an adult discrimination identifier based on an adult discrimination algorithm, and to authenticate an adult, and to sell an adult article such as cigarettes and alcohol to an authenticated adult. It is acceptable.

That is, the adult authentication device 120 does not need to separately register the face of the user to determine whether an adult is present, and the user only needs to face toward the image capturing unit 210 of the adult authentication device 120. .

4 and 5 are exemplary diagrams of a face detection algorithm according to an embodiment of the present invention.

The adult authentication device 120 generates the input image data using the image capturing unit 210, generates the multi-level resolution reduction data by gradually reducing the resolution of the input image data, and reduces each of the multi-level resolution reduction data. The image search data for each part searched at a predetermined pixel interval is generated by the search image window 410 every time, and whether the image search data for each part is a face image or a non-face image, and as a result of the check, the image search data for each part is checked. If is a face image, the image search data for each part is confirmed as a face image.

In addition, the adult authentication device 120 extracts an image pattern from the input image data, checks whether the image pattern corresponds to the face image using the face detection delimiter, and as a result, the image pattern corresponds to the face image. In this case, the input image data is confirmed as a face image.

Here, as the face detection algorithm, the Adaboost algorithm, which is one of the learning algorithms, may be used.

That is, the adult authentication device 120 may determine whether the image search data for each part is a face image or a non-face image using the Adaboost algorithm.

The Adaboost algorithm is a learning algorithm proposed by Freund. It is one of the boosting methods that constitutes one strong classifier with several weak classifiers.

In other words, the Adaboost algorithm basically proceeds with the probability distribution of the data. After learning a weak delimiter, the Adaboost algorithm lowers the probability distribution for correctly classified data and increases the probability distribution for incorrectly classified data. In addition, when learning weak delimiters, it is a way of learning by focusing on misclassified data.

Viola & Jones used the Adaboost algorithm for face detection, and in face detection, it consists of dozens of strong delimiters from thousands of weak delimiters. The face detection delimiter of the present invention refers to dozens of strong delimiters.

On the other hand, when the adult authentication device 120 assumes the input image data generated using the image capturing unit 210 as the background image shown in FIG. 4, the adult authentication device 120 determines the resolution of the input image data. Generates multi-stage reduced data that has been reduced in stages.

For example, if it is assumed that the resolution (size) of the input image data is 320x240 as shown in FIG. 4 and the size of the search image window 410 is 16x16, the adult authentication apparatus 120 may determine the resolution (the size of the input image data). Step by step (ie, 320x240-> 200x150-> 160x120…) to create multi-resolution reduced data. Here, the multi-stage resolution reduction data in which the resolution (size) is reduced in stages may be the same as the background image 160x120 of FIG. 5.

In addition, the adult authentication device 120 extracts an image pattern from the input image data, checks whether the image pattern corresponds to the face image using the face detection delimiter, and as a result, the image pattern corresponds to the face image. In this case, the input image data is confirmed as a face image.

For example, the adult authentication device 120 generates image search data for each part searched at predetermined pixel intervals for each reduction step of the multi-resolution reduction data by the corresponding size of the search image window 410. Partial image search data (ex: 16x16) is to determine whether or not the face image by the face detection identifier based on the Adaboost algorithm.

When the image search data for each part is determined as the face image, the size and position of the face image may be finally determined according to the size and the searched position of the input image data.

That is, the size of the search image window 410 is the same as 16x16, but the resolution (size) of the input image data is gradually reduced (that is, 320x240-> 200x150-> 160x120…) to search the generated multi-stage reduced data. When the input image data is 320x240, the image search data for each part searched by the search image window 410 is the same as that of FIG. 4, but the multi-stage resolution reduction data reduced to 160x120 is 16x16. When searching with the image window 410, the image corresponding to the face can be distinguished as shown in FIG. 5.

6 and 7 are exemplary diagrams of an image pattern according to an exemplary embodiment of the present invention.

The adult authentication device 120 extracts an image pattern from the input image data, checks whether the image pattern corresponds to the face image by using the face detection separator, and if the image pattern corresponds to the face image as a result of the checking, This function checks the input image data as a face image.

That is, the adult authentication device 120 detects a face image by analyzing only the image pattern in the entire area of the input image data.

For example, the image pattern for checking whether the image is a face is as shown in FIGS. 6A to 6D, and the adult authentication device 120 is shown in FIGS. 7A to 7C. As shown in the drawing, whether the image pattern of the input image data corresponds to the face image is determined.

8 is an exemplary view of an adult discrimination algorithm according to an embodiment of the present invention.

The adult authentication device 120 checks whether the face image is an adult face or a minor face using an adult discrimination identifier based on an adult discrimination algorithm.

Here, the SVM (Support Vector Machine) algorithm, which is one of learning algorithms, may be used as an adult identification algorithm.

That is, the face image is determined whether the face image is an adult face or a minor face using an adult discriminant discriminator based on the SVM algorithm.

The adult authentication apparatus 120 does not have a separate database for adult authentication, and may determine whether the adult face or a minor face is determined only by a predetermined adult discriminator. Since the database is needed only for the learning process for obtaining the adult discrimination or not separator, the processing time for the adult authentication device 120 to confirm whether the face image is an adult face or a minor face is very fast.

One of the learning algorithms, the SVM algorithm, is an algorithm that classifies two classes. That is, as shown in FIG. 8A, a boundary line that can be divided into class 1 and class 2 may be referred to as a separator.

That is, when the SVM algorithm is used for adult authentication, a plurality of face images of 0-18 years old are set to Class 1, and a plurality of face images of 19 years old or older are set to Class 2 to obtain one adult discriminator. will be.

The discriminator for discriminating an adult becomes probabilistic reference data that can determine which class of the face image belongs to class 1 or class 2.

However, since the data distribution of each class shows a high-dimensional nonlinear distribution, a nonlinear adult discriminator discriminator will be obtained as shown in FIG.

In the process of finding the adult discriminator using the SVM algorithm, data about eye coordinates, gender, facial expression, and facial contour according to age may be extracted and classified into class 1 and class 2.

Here, the number of face images aged 0-18 years and face images 19 years or older that are necessary to obtain an adult discriminator identifier for adult authentication based on the SVM algorithm is not particularly determined.

However, the larger the number of face images aged from 0 to 18 years old and the face images over 19 years of age needed to obtain the adult discrimination discriminator, the more accurate adult discrimination discriminator can be made.

On the other hand, the number of face images aged 0-18 years old and face images 19 years old or older that are necessary to obtain an adult discriminator is not specifically determined. Only the number of chapters will be enough to get an adult discriminator.

In other words, at least about 3000 adult face images and about 3000 minor face images are required to obtain an adult discriminator identifier based on the SVM algorithm. You will get a distinguishable adult discriminator.

In addition, in case the user appears to be in their 20s to 30s or in the case of 20s to 30s, the user needs to obtain various face images according to each age to obtain an adult discriminator.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

1 is a block diagram schematically illustrating an authentication system using face recognition according to an embodiment of the present invention;

2 is a block diagram schematically showing an adult authentication device according to an embodiment of the present invention;

3 is a flowchart illustrating an authentication method using face recognition according to an embodiment of the present invention;

4 and 5 are exemplary diagrams of a face detection algorithm according to an embodiment of the present invention;

6 and 7 are exemplary views of an image pattern according to an embodiment of the present invention;

8 is an exemplary view of an adult discrimination algorithm according to an embodiment of the present invention.

<Description of Symbols for Main Parts of Drawings>

110: subject 120: adult authentication device

210: image capturing unit 220: control unit

230: storage unit

Claims (10)

(a) receiving input image data; (b) checking whether the input image data is a face image or a non-face image by using a face detection identifier based on a face detection algorithm; (c) If the input image data is the face image as a result of checking in the step (b), whether the face image is an adult face or a minor face using an adult discrimination or discriminator based on an adult discrimination algorithm; Confirming; And (d) performing adult authentication when the face image is the adult face as a result of the checking in step (c) Authentication method using facial recognition, characterized in that it comprises a. The method of claim 1, After step (d) above, (e) generating multi-level resolution reduction data by gradually reducing the resolution of the input image data; generating image search data for each part searched at a predetermined pixel interval with a search image window for each reduction step of the multi-stage resolution reduction data; (g) checking whether the image search data for each part is the face image or the non-face image using the face detection delimiter; And (h) confirming the image search data for each part as the face image when the image search data for each part is the face image as a result of the checking in step (g). Authentication method using facial recognition, characterized in that it comprises a. The method of claim 2, After step (h), (i) extracting an image pattern from the input image data; (j) checking whether the image pattern corresponds to the face image using the face detection delimiter; And (k) confirming the input image data as the face image when the image pattern corresponds to the face image as a result of the checking in step (j) Authentication method using facial recognition, characterized in that it comprises a. The method of claim 1, The face detection algorithm, Authentication method using face recognition, characterized in that the Adaboost algorithm. The method of claim 1, The adult determination algorithm, Authentication method using face recognition, characterized in that the SVM (Support Vector Machine) algorithm. The method of claim 1, The face detection separator is, And a probabilistic reference data for distinguishing the face image or the non-face image. The method of claim 1, The adult discrimination discriminator, And a probabilistic reference data for distinguishing the adult face or the minor face. An image capturing unit obtaining an image photographing a subject; Generates input image data using the image capturing unit, and checks whether the input image data is a face image or a non-face image by using a face detection identifier based on a face detection algorithm. When the data is the face image, an adult discrimination or discriminator based on an adult discrimination algorithm is used to determine whether the face image is an adult face or a minor face. As a result of the check, the face image is the adult face. If, the control unit for controlling to perform the adult authentication; And A storage unit storing one or more pieces of information of the input image data, the face detection separator, the face image, and the adult discrimination separator; Authentication apparatus using face recognition, characterized in that. The method of claim 8, The control unit, Generating multi-step resolution reduction data in which the resolution of the input image data is reduced in stages, and generating image search data for each part searched at a predetermined pixel interval with a search image window for each reduction step of the multi-step reduction data; By using a face detection separator, it is determined whether the image search data for each part is the face image or the non-face image, and as a result of the check, when the image search data for each part is the face image, the image search data for each part is used. The adult authentication device using face recognition, characterized in that for checking the face image. The method of claim 8, The control unit, Extracting an image pattern from the input image data, and checking whether the image pattern corresponds to the face image by using the face detection separator, and if it is determined that the image pattern corresponds to the face image, The apparatus of claim 1, wherein the input image data is identified as the face image.

Images