KR20110012820A - Indivisual authentication apparatus according to the motion mixture of the face element and method thereof - Google Patents

Abstract

PURPOSE: An individual authentication apparatus according to the motion mixture of a face element, and a method thereof are provided to authenticate a face of a user by comparing a motion pattern with authentication data in face configuration components. CONSTITUTION: An image processing unit(11) processes an image of a camera unit(10), and a face region detection unit(12) detects a facial region from the processed image. A face component motion detection unit(13) detects a motion pattern of face configuration elements from the detected face region. A face authentication unit(14) compares the detected face region with the authentication data storage unit(15).

Description

{Indivisual authentication apparatus according to the motion mixture of the face element and method}

The present invention relates to a personal authentication device and a method for authenticating a person as a person when the motion combinations of the face components coincide.

Due to the explosive growth of modern information devices, computers and Internet users, the necessity of protecting personal data and restricting access is emerging, and various personal authentication methods have been proposed.

Among them, the authentication method using biometric information is a method of authenticating an individual based on fingerprint, iris, vein, and face information, which is unique to each individual, and prevents the risk of losing a password or a key and does not use an additional key. Due to its advantages, such as convenience, many researches are being actively conducted.

Among them, the facial recognition method using face information is based on the non-contact authentication characteristic that does not cause physical information access with the device when personal information is authenticated. It is highly preferred in terms of securing.

However, such a facial recognition system may be vulnerable to imitation of a subject by illegal photo information of the authenticator when authenticating based on face information. In order to suppress authentication by indiscriminate simulation of such a subject, an authentication method using three-dimensional face information has been proposed, but high performance system performance is generally required for real-time implementation of three-dimensional authentication processing.

In order to solve such a problem, a technique for blocking an authentication attempt based on false information such as a picture or a photograph simulated during facial recognition using face information is disclosed in Korean Laid-Open Patent No. 2009-0135188.

The above patent discloses determining one or more motion regions and motion directions where a face region is detected on a camera screen, and if there is a certain face movement in the detected face region, determines that it is a real face, otherwise it is determined as a photograph. .

If it is determined as a real face, it is determined whether the movement of the face is a physiological movement and distinguishes the final real face from the photograph.

Another conventional technique is disclosed in Korean Patent Laid-Open Publication No. 2006-0101259 to block authentication attempts by false information such as pictures or photos simulated during facial recognition using face information.

In the above patent, when the camera and the liquid crystal monitor are provided on the same surface, the liquid crystal monitor is necessarily reflected in the eyes of the user when taking a picture, and the reflected image reflected in the eye can be displayed on the liquid crystal monitor. Therefore, after changing the image of the monitor to take a face image, it is possible to determine the degree of color change of the image reflected in the vicinity of the pupil, and block it when an illegal authentication attempt is made by a photograph or the like.

Another conventional technology is Korean Patent Publication No. 2006-0101255, which suggests a method for blocking authentication by false two-dimensional information such as a photo by using a distance sensor when face authentication using a mobile phone.

In the above patent, after acquiring a face image to be authenticated, the distance to the face is measured by three distance sensors, and if there is a difference, it is displayed as a three-dimensional face. It suggests a technique for blocking.

However, the methods disclosed in the above patents can easily distinguish that the facial information to be authenticated is a photo or video information by detecting the movement of a specific part when performing facial recognition based on the face information of the user. For example, if an illegal user can arbitrarily synthesize and manipulate a specific part of an authenticator based on video information, the security may be seriously threatened by reacting promptly to a real-time motion request indicated by the system.

Accordingly, the present invention is to solve the above problems, and when the facial authentication based on the face information is detected, the movement order of the detected area after patterning the movement of the eyes, mouth, brow, etc. that is the authenticator face component Thus, the personal authentication device using a motion combination of facial components that distinguishes the authenticity of the face of the subject of authentication even when the image of the authenticator's face is randomly manipulated and displayed, as well as easily distinguishing the simulated information without movement. And a method thereof.

The present invention for achieving the above object, the authentication data storage unit for storing the facial authentication data; A camera unit for receiving an image; An image processor which processes and outputs an image input from the camera unit; A face region detector which receives an image processed by the image processor and detects and outputs a face region; A facial component motion detector for detecting a movement pattern of the facial component in the face region detected by the face region detector and determining whether it matches the stored pattern; And a face authentication unit configured to perform authentication by comparing the face area detected by the face area detection unit with authentication data stored in the authentication data storage unit.

In addition, the face area detection unit of the present invention is characterized by detecting the face area using the Adaboost learning algorithm.

In addition, the facial component detected by the facial component motion detection unit of the present invention is characterized in that it comprises at least one of the head, eyes, mouth.

In addition, the method of the present invention comprises the steps of (A) the image processing unit for processing the image input from the camera unit and outputs; (B) a face region detection unit receives an image processed by the image processor and detects and outputs a face region; (C) a face component motion detector detecting a movement pattern of a face component in the face region detected by the face region detector and checking whether the face component motion detector matches the stored face movement pattern; And (D) a face authentication unit comparing the face area detected by the face area detection unit with authentication data stored in the authentication data storage unit to perform authentication.

In the method of the present invention, in the step (B), the face area detection by the face area detection unit is characterized in that the face area is detected using an Adaboost learning algorithm.

In the step (C) of the method of the present invention, the facial component motion detection may be performed by using the difference in brightness of the facial component between the current frame and the previous frame.

In the step (C) of the method of the present invention, the facial component motion detection may be performed by using a change amount of the motion vector amount of the facial component between the current frame and the previous frame.

According to the present invention as described above, by applying a step of determining and matching the specific movement order for the face components after the face detection, real-time authentication attempts through video face information and random video manipulation and synthesis of the authenticator photographed as a video In addition, since the system can be blocked from access, the reliability of the system can be improved.

In addition, according to the present invention, when facial authentication is performed through face recognition, the movement of facial components such as eyes and mouth may be determined, and thus authentication attempts based on simulated data such as photographs and statues may be blocked.

Further, according to the present invention, unlike the prior art in which additional elements such as a heat sensor, a distance sensor, or a luminaire are required to determine an actual user, the present invention can be applied to an existing face recognition system by only a software upgrade without adding hardware.

In addition, according to the present invention, since the change value is detected by comparing the frame-to-frame only in the detected region of interest, not the entire region of the screen, motion detection can be performed in real time with minimal system load.

In addition, according to the present invention, since it is based on the motion determination, it is possible to apply the face authentication system regardless of the characteristics of the two-dimensional and three-dimensional facial information.

Hereinafter, a personal authentication device and a method thereof based on motion combinations of facial components according to a preferred embodiment of the present invention will be described in detail with reference to FIG. 1.

1 is a block diagram illustrating a schematic configuration of a personal authentication device using a motion combination of facial components according to an embodiment of the present invention.

Referring to FIG. 1, a personal authentication apparatus using a combination of motions of facial components according to an exemplary embodiment of the present invention may include a camera unit 10, an image processor 11, a face region detector 12, and a facial component. The motion recognition unit 13 includes a face recognition unit 14, an authentication data storage unit 15, and a display unit 16.

Here, the camera unit 10 receives and processes a real-time image as a sensor input, thereby outputting the image to be processed in units of frames.

In addition, the image processor 11 basically changes the original (RAW) data received through the camera unit 10 into data in the form of RGB and YUV capable of image processing, and at the same time AE (Auto Exposure), AWB ( It handles tasks such as Auto White Balance (NR) and Noise Reduction (NR).

Next, the face area detection unit 12 detects and outputs a face area. The method used for detecting the face area includes a knowledge based method, a feature based method, a template matching method, an appearance based method, and a face color information method. , Time information, and the like.

Then, the facial component motion recognition unit 13 recognizes the movement occurring in the face motion detection region 21 of FIG.

For example, it is determined whether the movement of the brow 21, the eye 22, the mouth 23, and the like occurs, and not only the transmission of the determination result to the image processor 11 but also the movement stored in the memory (not shown). The degree of agreement between the specific pattern and the detected movement pattern is determined.

On the other hand, the face authentication unit 14 compares the face authentication data previously stored in the authentication data storage unit 15 with the face image input through the camera to determine whether to authenticate.

The operation of the personal authentication device based on the motion combination of the facial components configured as described above will now be described in detail.

First, the camera unit 10 receives and processes a real-time image as a sensor input, and outputs the image to be processed in units of frames.

Then, the image processing unit 11 basically changes the raw (RAW) data received through the camera unit 10 into data in the form of RGB and YUV capable of image processing, and simultaneously AE (Auto Exposure) and AWB (Auto White Balance). ), And processes such as noise reduction (NR).

In addition, the image processor 11 causes the face region detector 12 to detect the face region of the input screen, and when the face region is detected by the face region detector 12, the image region 11 moves the facial component information. Transfer to the recognition unit 13, to perform the motion recognition.

When the face motion is successfully recognized by the face component motion recognition unit 13, the image processor 11 transmits the face area to the face authentication unit 14 to store the authentication data stored in the authentication data storage unit 15. The face authenticator 14 may determine whether to authenticate by comparing the data.

Then, the face authentication unit 14 compares the face authentication data stored in the authentication data storage unit 15 with the face image input through the camera, determines whether to be authenticated, and outputs the determination result to the image processing unit 11. .

Finally, the image processor 11 notifies the user of the authentication result through the display unit 16.

3 is a flowchart of a personal authentication method using a motion combination of facial components according to a preferred embodiment of the present invention.

First, the face area detection unit detects a face by searching for a face in the input image (S100).

The face detection unit may detect a face using various methods. For example, the face region detection unit may detect a face by using an Adaboost learning algorithm and matching the face learning data stored therein with an input image. The size information is extracted to detect a face.

Subsequently, the face position and size information provided to the image processing unit is provided to the face component motion recognition unit, and the face component motion recognition unit detects a face component (eye, nose, mouth, etc.) based on the detected face information. In operation S101, the face composition regions 22, 23, 24, 25, 26, and 27 are detected based on the center position and the size information of the face.

Next, the face component motion recognition unit determines whether there is motion in the detection area (S102), and when it moves, determines whether or not it matches the stored motion pattern (S103).

The determination of motion in the detection area performed by the face component motion recognition unit can be obtained through the motion vector size (motion vector amount change) or the difference image component (brightness difference) between the current frame and the previous frame. The noise and the size of the face movement of the authenticator are taken into consideration.

The facial component motion recognition unit judges that there is motion only when it is equal to or greater than a predetermined threshold value, and in this case, it is insensitive to ambient noise and can perform stable motion detection specialized for motion to the authenticator.

In addition, the coincidence operation with the motion pattern performed by the face component motion recognition unit may improve security for authentication by performing a coincidence operation with a specific pre-input motion pattern as well as simple motion detection.

For example, when patterning with the movement of the eye and mouth regions as shown in FIG. 4, the right eye 23, the left eye 24, the right eye and the mouth 23. Arbitrary pattern combinations such as mouth movements 24 + 25 may be used as additional authentication information.

If the distance of the authenticator's face from the camera is kept constant, the moving part can also be differentiated according to the authenticator's face size and appearance, so that a reliable security effect is possible.

After determining whether the motion matches the movement pattern, an authentication operation is performed between the face area detected by the face authentication unit through the face area detection unit and the data stored in the authentication data storage unit (S104).

By comparing the unique feature amount between the unique feature amount and the stored data from the detected face region data, if the difference in feature amount is within a predetermined predetermined threshold, it is determined to be the same person (S105) otherwise, it is determined to be the same person. (S106).

The intrinsic feature amount of the face is linearly determined by principal component mode analysis (PCA) of pixel brightness, eigenface, fisherface, Haae feature vector, Gabor feature vector, or Gabor feature vector. Acquisition is based on techniques such as vector operations obtained by linear discriminant analysis.

1 is a block diagram showing a schematic configuration of an apparatus for personal authentication by motion combination of facial components according to an embodiment of the present invention.

FIG. 2 is an exemplary view illustrating a facial component used in the facial component motion recognition unit of FIG. 1. FIG.

3 is a flow chart of a personal authentication method by motion combination of facial components according to a preferred embodiment of the present invention.

4 is an exemplary view showing a facial component motion pattern used in the present invention.

<Explanation of symbols for the main parts of the drawings>

10: camera unit 11: image processing unit

12: face region detection unit 13: face component motion recognition unit

14: face authentication unit 15: authentication data storage unit

Claims (7)

An authentication data storage unit for storing facial authentication data; A camera unit for receiving an image; An image processor which processes and outputs an image input from the camera unit; A face region detector which receives an image processed by the image processor and detects and outputs a face region; A facial component motion detector for detecting a movement pattern of the facial component in the face region detected by the face region detector and determining whether it matches the stored pattern; And And a face authentication unit for performing authentication by comparing the face area detected by the face area detection unit with the authentication data stored in the authentication data storage unit. In claim 1 The face area detection unit detects a face area using an Adaboost learning algorithm. In claim 1 The facial component component detected by the facial component motion detector includes at least one of a glans, an eye, and a mouth. (A) an image processor processing and outputting an image input from the camera unit; (B) a face region detection unit receives an image processed by the image processor and detects and outputs a face region; (C) a face component motion detector detecting a movement pattern of a face component in the face region detected by the face region detector and checking whether the face component motion detector matches the stored face movement pattern; And And (D) a face authentication unit comparing the face area detected by the face area detection unit with the authentication data stored in the authentication data storage unit to perform authentication. In claim 4 In step (B), the face area detection by the face area detection unit detects a face area using an Adaboost learning algorithm. In claim 4 In the step (C), the facial component motion detection is performed through the difference in brightness of the facial component region between the current frame and the previous frame. In claim 4 In step (C), the facial component motion detection is performed through the motion vector variation of the facial component region between the current frame and the previous frame.

Images