KR101897072B1 - Method and apparatus for verifying facial liveness in mobile terminal - Google Patents

Abstract

According to the present invention, a facial liveness verification method comprises the steps of: acquiring M reference face images at different angles in advance through a camera; forming reference triangles from the acquired first reference face image and the acquired M^th reference face image; acquiring a first real-time face image and an N^th real-time face image, which are successive at different angles, through the camera in real time; forming real-time triangles from the acquired first real-time face image and the acquired N^th real-time face image; extracting a reference triangle most similar to the real-time triangle among the generated reference triangles by comparing the similarity of the reference triangles generated for each of the M reference face images to a real time triangle generated for each real time face image obtained through the camera in real time; determining a face posture of a reference face image corresponding to the most similar reference triangle as a face posture of a real-time image obtained in real time; and determining whether a real-time face image is a forged face image or a living person′s face image based on a change in the face posture of the determined real-time image. According to the present invention, it is possible to accurately determine whether a face image is forged in a terminal.

Description

[0001] METHOD AND APPARATUS FOR VERIFYING FACIAL LIVENESS IN MOBILE TERMINAL [0002]

The present invention relates to a method and apparatus for verifying facial liveness in a mobile terminal.

2. Description of the Related Art Various types of user authentication methods have been proposed to improve the security of each electronic device with the increasing use of electronic devices capable of displaying and reproducing various types of contents such as smart phones and PCs.

The user authentication method includes a method of recognizing a slide pattern displayed on a screen when a user has a touch screen, a PIN method of inputting numbers, and a password input method of combining characters. In recent years, ) Authentication method is used.

Among them, face recognition technology is a technique of log-in to an electronic device (eg, a computer, a notebook, a mobile terminal and others) through facial recognition instead of inputting a password. Stored faces are compared with each other to permit user authentication for accessing the electronic device, that is, login.

However, the conventional face recognition technology mainly focuses on the login of the electronic device. In order to restrict the use of the electronic device or to control the information access by verifying the authenticated fake user through the forgery of the face image of the user, I am not interested. Therefore, there is an urgent need to develop a technique for verifying a forgery face image that can solve such a problem.

SUMMARY OF THE INVENTION The present invention is conceived to solve the above-mentioned problems, and it is an object of the present invention to provide a facial live recording system for determining whether a captured image is forged or not by using a triangle generated by connecting end points of face landmarks included in real- And to provide a verification method and verification device.

 Other objects and advantages of the present invention will become apparent from the following description, and it will be understood by those skilled in the art that the present invention is not limited thereto. It will also be readily apparent that the objects and advantages of the invention may be realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims.

According to an aspect of the present invention, there is provided a method for verifying facial liveness in a facial-liveness verification apparatus, comprising: acquiring M reference face images from a different angle in advance through a camera; The method comprising the steps of: forming reference triangles from an M-th reference face image obtained at different angles; acquiring real-time first real-time face images and N-th real- And generating real-time triangles from the obtained N-th real-time facial images, and generating real-time triangles based on the reference triangles generated for each of the M reference-face images and the real-time triangles generated for each real- By comparing the real triangles of the generated reference triangles with Determining a face posture of a reference face image corresponding to the most similar reference triangle as a face posture of a real-time image obtained in real time; And determining whether the real-time face image is a forged face image or a living person's face image based on the determined change in the face posture of the real-time image.

The step of acquiring the first real-time face image and the Nth real-time face image successively from different angles through the camera in real time may include a step of instructing the face moving direction through the speaker, And realizing the first real-time face image and the N-th real-time face image through the camera in succession as the user moves the face at different angles according to the image display for moving the face The step of determining whether the face image is a forgery-altered face image or a live person's face image based on a change in the face posture of the image includes: determining whether a change in the face posture of the determined real- To the direction of movement of the image displayed in And determining whether the real time face image is a forged face image or a living person's face image based on whether or not the face image changes in accordance with the image display for moving the face.

Wherein the step of forming reference triangles from the obtained first reference face image and the obtained M reference face image comprises the steps of: extracting a face landmark for each of the obtained M reference face images; And forming a plurality of triangles by connecting the end points of the extracted face landmarks to the reference triangles to generate the reference triangles.

The face liveness verification method includes a ratio of a first width between a width of a first triangle and a width of a second triangle contained in a first real time face image and a width of a third triangle included in the Nth real time face image, Calculating a second width ratio between the widths of the first real face image and the Nth real face when the difference between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, Judging the image as a forged face image, and when the difference value between the first and second width ratios is larger than the reference width ratio difference value, the first real-time face image and the N-th real- A ratio change criterion determining step of determining a ratio change criterion; And determining that the real-time face image is a living face image of a living person only if the real-time face image is determined to be a living face image of a living person in both of the posture criterion determination step and the rate change criterion determination step have.

The step of forming a real-time triangle comprises: extracting a face landmark from the obtained first real-time face image and the obtained N real-time face image; And forming triangles by connecting the obtained first real-time face image and end points of a face landmark extracted from the obtained N-th real-time face image.

Wherein the face realness verification method comprises: calculating a width of a first triangle and an area of a second triangle included in the first real-time face image; calculating a width of the first triangle and the second triangle included in the first real- Extracting a third triangle and a fourth triangle included in the Nth real-time face image corresponding to the first real face image, calculating a width of the third triangle included in the Nth real- Calculating a ratio of a first width between the step and the width of the first triangle and the width of the second triangle and a second width ratio between the width of the third triangle and the width of the fourth triangle.

The first triangle and the second triangle may be triangles sharing one side.

The face liveness verifying apparatus according to another embodiment of the present invention includes a first reference face image and an M reference face image at different angles, and a first real-time face image and an N-th real- Wherein the first real-time face image is obtained from the obtained first real-time face image and the obtained M real-time face image, A triangle generating unit for generating real triangles and comparing the similarities of the real triangles generated for each of the M reference face images and the reference triangles generated for each real face image to generate a reference triangle A triangle extracting unit for extracting the most similar reference Posture determination unit for determining the posture of the face based on the face image corresponding to the rectangular faces to the position of the real-time image obtained in real time; And a liveness determination unit for determining whether a real-time face image is a forged face image or a live person's face image based on a change in the face posture of the determined real-time image.

The camera can continuously display the first real time image in accordance with the user's face moving at different angles according to the instruction voice for the face movement direction through the speaker or the image display for moving the face according to the moving direction of the image displayed on the display unit, And the liveness determination unit obtains the face image and the Nth real time face image in real time, and the liveness determination unit determines that the change in the face posture of the determined real time image is in the direction of the face movement direction through the speaker, It is possible to judge whether the real-time face image is a forged face image or whether it is a living person's face image based on the posture criterion based on whether or not the real-time face image is changed in accordance with the image display for moving the face.

The face loudness verifying apparatus further includes a landmark extracting unit for extracting a face landmark for each of the M reference face images, and the triangle generating unit extracts a face landmark extracted for each of the M reference face images landmarks of the reference triangles may be connected to form a plurality of triangles to generate reference triangles.

Wherein the face liveness verifying apparatus includes a first width ratio between the width of the first triangle and the width of the second triangle included in the first real time face image and a width of the third triangle included in the N real time face image, Wherein the liveness determining unit determines that the first real time ratio and the second real time ratio are smaller than the reference width ratio difference value when the difference between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, The face image and the N-th real time face image are judged as a forged face image, and when the difference value between the first width ratio and the second width ratio is larger than the reference width ratio difference value, The N-th real-time face image is judged as the face image of a living person, It is possible to judge whether the image is unknown or whether it is a face image of a living person.

The liveness judgment unit can judge that the real-time face image is a living face image of a living person only if the real-time face image is judged to be a living face image of a living person, .

Wherein the face presence verifying apparatus further comprises a landmark extracting unit for extracting a face landmark from the obtained first real time face image and the obtained Nth real time face image, 1 real-time face image and the end points of a face landmark extracted from the obtained N-th real-time face image to form triangles.

The face liveness verifying apparatus includes a triangle extracting unit for extracting a third triangle and a fourth triangle included in the first real-time face image and the first real-time face image corresponding to the first real- Wherein the calculation unit calculates the width of the first triangle and the width of the second triangle included in the first real-time face image, and calculates the width of the third triangle included in the N-th real- The width of the triangle is calculated and the ratio of the first width between the width of the first triangle and the width of the second triangle and the width of the third triangle and the second width ratio between the width of the fourth triangle can be calculated.

The first triangle and the second triangle may be triangles sharing one side.

According to the disclosed invention, whether the obtained image is forged or not is determined using a triangle generated by connecting end points of face landmarks included in images obtained in real time, thereby accurately determining whether the terminal image is forged or falsified .

1 is a diagram for explaining a face presence verification apparatus.
FIGS. 2 to 5 are diagrams for explaining a face liveness verification method.
FIG. 6 is a flowchart illustrating a method for verifying facenessness of a face based on a face posture standard according to an exemplary embodiment of the present invention. Referring to FIG.
FIG. 7 is a flowchart illustrating a method for verifying facenessness in which a livenessness is determined based on a ratio change criterion according to an embodiment of the present invention.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram for explaining a face presence verification apparatus.

Referring to FIG. 1, the face presence verification apparatus 100 includes a camera 110, a landmark extraction unit 120, a triangle generation unit 130, a triangle extraction unit 140, an operation unit 150, And a posture judging unit 170. The posture judging unit 170 judges whether or not the posture is acceptable.

The camera 110 can acquire real-time first real-time face images and Nth real-time face images that are continuous at different angles. For example, the face image may be an image obtained at different angles such as a front face image, a left side face image, a right side face image, an upper side face image, a lower side face image, and the like.

The camera 110 may display an instruction voice for a face movement direction through a speaker or an image display for moving a face in accordance with the moving direction of the image displayed on the display unit, The first real time face image and the Nth real time face image can be obtained in real time. For example, the instruction voice may be a voice that can direct the direction of the user's facial movement such as "rotate the face to the left "," raise the face to the upper side ". Also, for example, the image display may be an image displayed on the display unit ('donut, mask, etc.') and may be displayed on the display unit to move the image to the left so that the user can move the face to the left.

The camera 110 may obtain M reference face images at different angles. The reference face image is an image obtained before obtaining the real time face image, which means an image obtained in advance to set a specific reference.

The landmark extracting unit 120 may extract a face landmark from the obtained first real-time face image and the obtained N-th real-time face image.

The landmark extracting unit 120 may extract a landmark for each of the M reference face images.

The triangle generator 130 may form triangles from the obtained first real-time face image and the obtained N real-time face image.

For example, the triangle generator 130 may form triangles by connecting the obtained first real-time face image and the end points of a face landmark extracted from the obtained N real-time face image.

As another example, the triangle generator 130 may generate a plurality of triangles by connecting the end points of the face landmarks extracted for each of the M reference face images to generate a reference triangle.

The triangle extracting unit 140 extracts the first triangle and the second triangle included in the first real-time face image among the triangles formed by the triangle generating unit 130, and outputs the N-th real time The third triangle included in the face image, and the fourth triangle.

Here, the first triangle and the second triangle may be triangles sharing one side. This means that for two triangles sharing one side, the ratio of the width of the two triangles in the first 2D image and the ratio of the widths of the two triangles in the second 2D image (the image obtained continuously or at different angles) The first 2D image and the second 2D image can both be judged as 2D images.

The triangle extracting unit 140 compares the similarities of the reference triangles generated for each of the M reference face images and the triangles generated for each real-time face image, and extracts the reference triangle most similar to the real-time triangle among the generated reference triangles .

The arithmetic unit 150 calculates a ratio of a first width ratio between the width of the first triangle and the width of the second triangle included in the first real time face image and a width of the third triangle included in the N real time face image, The second area ratio between the widths of the first area and the second area can be calculated.

For example, the operation unit 150 may calculate the width of the first triangle and the width of the second triangle included in the first real-time face image.

The operation unit 150 may calculate the width of the third triangle included in the Nth real time face image and the width of the fourth triangle.

The operation unit 150 can calculate the ratio of the first width between the width of the first triangle and the width of the second triangle, and the second width ratio between the width of the third triangle and the width of the fourth triangle.

The liveness judging unit 160 can judge whether the real time face image is a forged face image or a live face image based on the change of the face posture of the real time image judged by the posture judging unit 170. [ For example, the liveness determination unit 160 determines whether the change in the face posture of the determined real-time image is an instruction voice for the face movement direction through the speaker or an image for moving the face in accordance with the movement direction of the image displayed on the display unit It is possible to determine whether the real time face image is a forged face image or whether it is a living person's face image based on the posture reference. Specifically, the liveness judging unit 160 judges whether the face of the user has moved in the direction coinciding with the voice ('rotate the face to the left') on the basis of the face posture, It is possible to judge that the face image of the user.

The liveness judging unit 160 judges the first real time face image and the Nth real time face image as a living person's face image when the difference value between the first width ratio and the second width ratio is larger than the reference width ratio difference value . The large difference in width ratio means that the first real-time face image and the Nth real-time face image are not 2D images ('3D images'). Accordingly, the liveness judging unit 160 does not input the first real-time face image and the Nth real-time face image through the camera while the user rotates / moves the 2D image, and inputs the face image through the camera as the live person moves As shown in FIG.

On the other hand, if the difference value between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, the liveness judging unit 160 sets the first real time face image and the Nth real time face image as a forged It can be judged. The small difference value of the width ratio means that the first real time face image and the Nth real time face image are 2D images. Accordingly, the liveness judging unit 160 judges that the user inputs the first real-time face image and the Nth real-time face image through the camera while rotating / moving the 2D image, and the currently obtained real- You can judge by image.

Only when the real-time face image is judged as a face image of a living person, the liveness judgment unit 160 judges whether the real-time face image is a living face image of a living person It can be judged. Thus, by judging the liveness through the two steps, the liveness determiner 160 according to the present invention can more accurately determine the liveness.

The posture determination unit 170 can determine the face posture of the reference face image corresponding to the most similar reference triangle as the face posture of the real-time image obtained in real time.

FIGS. 2 to 5 are diagrams for explaining a face liveness verification method.

2 to 5, in the face presence verification method, the face liveness verification apparatus can acquire the first real-time face image and the Nth real-time face image successive from different angles in real time through the camera . The first real-time face image, and the N-th real-time face image, the number of real-time face images obtained in real-time can be variously set.

Referring to FIG. 2, the face presence verification apparatus can extract a face landmark from the obtained first real-time face image and the obtained N-th real-time face image.

Referring to FIG. 3, the face-life verifying apparatus may form triangles from the obtained first real-time face image and the obtained N-th real-time face image. For example, the face-life verifying apparatus may form triangles by connecting the obtained first real-time face image and the end points of the face landmark extracted from the obtained N-th real-time face image.

Referring to FIG. 4, the face liveness verifier can calculate the width of the first triangle 400 included in the first real-time face image and the width of the second triangle 410. The first triangle 400 and the second triangle 410 are triangles that share one side.

The facial verity verification apparatus includes a first triangle 400 included in the first real-time facial image, and a third triangle 420 and a fourth triangle 420 included in the N-th real-time facial image corresponding to the second triangle 410 Can be extracted. For example, the third triangle 420 is a triangle corresponding to the first triangle 400, the fourth triangle 430 is a triangle corresponding to the second triangle 410, Can be recognized.

The face presence verifying apparatus can calculate the width of the third triangle 420 and the width of the fourth triangle 430 included in the Nth real time face image.

Wherein the face liveness verifying apparatus includes a first width ratio between the width of the first triangle and the width of the second triangle included in the first real time face image and a width of the third triangle included in the N real time face image, The second width ratio between the widths of the triangle can be calculated.

The face liveness verifying apparatus can determine the first real time face image and the Nth real time face image as a forged face image when the difference value between the first width ratio and the second width ratio is smaller than the reference width ratio difference value .

The face liveness verifying apparatus can determine the first real time face image and the Nth real time face image as a living person's face image when the difference value between the first width ratio and the second width ratio is larger than the reference width ratio difference value .

The face liveness verifying apparatus can acquire M reference face images from the different angles in advance through the camera.

The face presence verifying apparatus can extract a face landmark for each of the M reference face images obtained.

Referring to FIG. 5, the face liveness verification apparatus may generate a plurality of triangles by connecting the end points of the face landmarks extracted for each of the M reference face images to generate a reference triangle. For example, the M reference face images may include a left gaze face image, a front gaze face image, a right gaze face image, an upper gaze face image, and the like.

The face liveness verifying apparatus compares the similarity of the reference triangles generated for each of the M reference face images and the real-time triangles generated for each real-time face image obtained through the camera in real time, The reference triangle can be extracted.

The facial verity verification device judges the face posture of the reference face image corresponding to the most similar reference triangle as the face posture of the real-time image obtained in real time, and determines whether the real- It is possible to judge whether the face image is a face image of a living person or a face image of a living person.

FIG. 6 is a flowchart illustrating a method for verifying facenessness of a face based on a face posture standard according to an exemplary embodiment of the present invention. Referring to FIG.

Referring to FIGS. 1 and 6, the face presence verifying apparatus acquires M reference face images from cameras at different angles in advance (500). The face liveness verifying apparatus forms reference triangles from the obtained first reference face image and the obtained M th reference face image (510).

The face liveness verifying apparatus compares the similarity of the reference triangles generated for each of the M reference face images and the real-time triangles generated for each real-time face image obtained through the camera in real time, A reference triangle is extracted (520).

The facial verity verification apparatus judges the face posture of the reference face image corresponding to the most similar reference triangle as the face posture of the real-time image obtained in real time (530).

The facial verity verification apparatus can determine whether the real-time face image is a forged face image or a living person's face image based on a change in the face posture of the judged real-time image (540).

For example, the face presence verifying apparatus may be configured to display an instruction voice for a face movement direction through a speaker or an image display for moving a face in accordance with a moving direction of an image displayed on the display unit, The first real-time face image and the N-th real-time face image can be continuously acquired through the camera in real time according to the movement.

Then, the face presence verifying apparatus determines whether the change of the face posture of the determined real-time image is an instruction voice for the face movement direction through the speaker or an image display for moving the face according to the moving direction of the image displayed on the display unit It is possible to judge whether the real-time face image is a forged face image or a living person's face image based on whether or not the face image is coincidentally changed.

As described above, the face liveness verifying apparatus determines the posture of the real-time face image based on the similarity degree of the triangle and determines whether or not the face image of the living person is based on the determined posture. It is possible to more accurately determine the posture of the face image and to judge the loudness based on the posture of the face image, thereby making it possible to more accurately determine the loudness.

FIG. 7 is a flowchart illustrating a method for verifying facenessness in which a livenessness is determined based on a ratio change criterion according to an embodiment of the present invention.

Referring to FIGS. 1 and 7, the facial-liveness verification apparatus acquires a first real-time face image and a Nth real-time face image successively from different angles through a camera in real time (600).

The face liveness verifying apparatus acquires 610 the first real-time face image and the Nth real-time face image successively from different angles in real time through the camera.

The face presence verification apparatus forms real-time triangles from the acquired first real-time face image and the obtained N-th real-time face image (620).

The facial verity verification apparatus calculates 630 the width of the first triangle and the width of the second triangle included in the first real-time face image.

The facial verity verification apparatus extracts a first triangle included in the first real-time face image and a third triangle and a fourth triangle included in the N-th real-time face image corresponding to the second triangle (640).

The facial-liveness verification apparatus calculates 650 the width of the third triangle and the width of the fourth triangle included in the Nth real-time face image.

The facial-liveness verification apparatus calculates 660 the ratio of the first width between the width of the first triangle and the width of the second triangle, and the second width ratio between the width of the third triangle and the width of the fourth triangle.

The face presence verifying apparatus determines whether the difference value between the first and second width ratios is smaller than the reference width ratio difference value (670).

If the difference between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, the face liveness verifying apparatus determines the first real-time face image and the Nth real-time face image as the forged face image (680) .

The facial verity verification apparatus judges that the first real time face image and the Nth real time face image are a living person's face image when the difference value between the first width ratio and the second width ratio is larger than the reference width ratio difference value .

The face liveness verifying apparatus may be configured such that only when the result of the determination based on the posture reference described in Fig. 6 and the result of judging based on the ratio change criterion shown in Fig. 7 are all the face images of the living person, It can be judged by the face image of a person.

The embodiments described may be constructed by selectively combining all or a part of each embodiment so that various modifications can be made.

It should also be noted that the embodiments are for explanation purposes only, and not for the purpose of limitation. In addition, it will be understood by those of ordinary skill in the art that various embodiments are possible within the scope of the technical idea of the present invention.

100: Facial loudness verifying device
110: camera
120: landmark extracting unit
130: triangle generator
140: triangle extraction unit
150:
160:
170:

Claims (15)

A method for verifying facial liveness in a facial verity verification apparatus,
Acquiring M reference face images from cameras at different angles in advance;
Forming reference triangles from the obtained first reference face image and the obtained M reference face image;
Acquiring a first real-time face image and an N-th real-time face image successively from different angles through a camera in real time;
Forming real-time triangles from the obtained first real-time face image and the obtained N-real-time face image;
The reference triangles generated for each of the M reference face images and the real time triangles generated for each real time face image obtained through the camera in real time are compared to extract the reference triangle most similar to the real time triangle among the generated reference triangles step;
Determining a face posture of the reference face image corresponding to the most similar reference triangle as a face posture of a real-time image obtained in real time; And
And determining whether a real-time face image is a forged face image or a living person's face image based on the determined change in the face posture of the real-time image.
The method according to claim 1,
Wherein acquiring the first real-time face image and the Nth real-time face image successively from the different angles through the camera in real time
According to the instruction voice for the face movement direction through the speaker or the image display for moving the face in accordance with the moving direction of the image displayed on the display unit, the user continuously moves the first real time face image And acquiring the N-th real time face image through the camera in real time,
The step of determining whether a forged face image or a living face image of a living person is based on a change in the face posture of the determined real-
Wherein the change of the face posture of the determined real-time image changes according to an instruction voice for the face movement direction through the speaker or an image display for moving the face according to the moving direction of the image displayed on the display unit, And a posture criterion judgment step of judging whether the real time face image is a forged face image or a live person's face image.
The method according to claim 1,
Wherein forming the reference triangles from the obtained first reference face image and the obtained M reference face image comprises:
Extracting a face landmark for each of the M reference face images; And
And creating a plurality of triangles by connecting the end points of the landmarks extracted for each of the M reference face images to generate a reference triangle.
3. The method of claim 2,
A first width ratio between the width of the first triangle and the width of the second triangle included in the first real time face image and a second width between the width of the third triangle included in the N real time face image and the width of the fourth triangle, Calculating a ratio;
Judging the first real-time face image and the N-th real-time face image as a forged face image when the difference value between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, A ratio change criterion determining step of determining the first real-time face image and the N-th real-time face image as a face image of a living person when the difference between the ratio and the second width ratio is larger than the reference width ratio difference value; And
And determining that the real-time face image is a living face image of a living person only if the real-time face image is determined to be a living face image of a living person in both of the posture criterion determination step and the rate change criterion determination step How to verify Liveness.
The method according to claim 1,
The step of forming the real-
Extracting a face landmark from the obtained first real-time face image and the obtained N-th real-time face image; And
And forming triangles by connecting the obtained first real-time face image and end points of a face landmark extracted from the obtained N-th real-time face image.
5. The method of claim 4,
Calculating a width of the first triangle and an area of the second triangle included in the first real time face image;
Extracting a third triangle and a fourth triangle included in the N th real time face image corresponding to the first triangle and the second triangle included in the first real time face image;
Calculating an area of a third triangle included in the Nth real time face image and an area of the fourth triangle; And
Calculating a ratio of a first width between the width of the first triangle and the width of the second triangle and a second width ratio between the width of the third triangle and the width of the fourth triangle, .
5. The method of claim 4,
Wherein the first triangle and the second triangle are arranged such that,
A method for verifying facial liveness, the triangles sharing one side.
A camera for acquiring in real time a first reference face image and an Mth reference face image at different angles, and a first real-time face image and a Nth real-time face image successive from different angles;
A third real-time facial image forming unit configured to form reference triangles from the obtained first real-time facial image and the obtained M real-time facial image, form real-time triangles from the obtained first real- ;
A triangle extracting unit for comparing the similarities of the reference triangles generated for each of the M reference face images and the real time triangles generated for each real time face image and extracting a reference triangle most similar to the real time triangle among the generated reference triangles;
A posture judging unit for judging the face posture of the reference face image corresponding to the most similar reference triangle as the face posture of the real time image obtained in real time; And
And a liveness determination unit for determining whether a real-time face image is a forged face image or a living person's face image based on a change in the face posture of the determined real-time image.
9. The method of claim 8,
The camera comprises:
According to the instruction voice for the face movement direction through the speaker or the image display for moving the face in accordance with the moving direction of the image displayed on the display unit, the user continuously moves the first real time face image And an Nth real time face image in real time,
Wherein the liveness judging unit judges,
Wherein the change of the face posture of the determined real-time image changes according to an instruction voice for the face movement direction through the speaker or an image display for moving the face according to the moving direction of the image displayed on the display unit, A facial liveness verifying apparatus for judging whether a real-time face image is a forged face image or a living person's face image based on an attitude reference.
9. The method of claim 8,
And a landmark extracting unit for extracting a face landmark for each of the obtained M reference face images,
The triangle generation unit may generate,
And connecting the end points of the face landmarks extracted for each of the M reference face images to form a plurality of triangles to generate a reference triangle.
10. The method of claim 9,
A first width ratio between the width of the first triangle and the width of the second triangle included in the first real time face image and a second width between the width of the third triangle included in the N real time face image and the width of the fourth triangle, And an operation unit for calculating a ratio,
Wherein the liveness judging unit judges,
Judging the first real-time face image and the N-th real-time face image as a forged face image when the difference value between the first width ratio and the second width ratio is smaller than the reference width ratio difference value, And the second width ratio is greater than the reference width ratio difference value, the real-time face image is obtained by using the ratio change reference for determining the first real-time face image and the N-th real- A facial liveness verifying apparatus for judging whether a forged face image or a living person's face image is an image.
12. The method of claim 11,
Wherein the liveness judging unit judges,
The face realization method judges the real-time face image as the face image of the living person only if the real-time face image is judged to be the living face image of the living person, Device.
9. The method of claim 8,
And a landmark extracting unit for extracting a face landmark from the obtained first real-time face image and the obtained N-th real-time face image,
The triangle generation unit may generate,
And forms triangles by connecting the obtained first real-time face image and end points of a face landmark extracted from the obtained N-th real-time face image.
12. The method of claim 11,
And a triangle extraction unit extracting a third triangle and a fourth triangle included in the N-th real time face image corresponding to the first triangle and the second triangle included in the first real time face image,
The operation unit,
Calculating a first triangle width and a second triangle width included in the first real time face image,
Calculating an area of the third triangle included in the Nth real time face image and an area of the fourth triangle,
A first width ratio between the width of the first triangle and the width of the second triangle, and a second width ratio between the width of the third triangle and the width of the fourth triangle.
12. The method of claim 11,
Wherein the first triangle and the second triangle are arranged such that,
A facial liveness verification device, which is a triangle sharing one side.

Images