KR100706871B1 - Method for truth or falsehood judgement of monitoring face image - Google Patents

Abstract

The present invention relates to a method for distinguishing the authenticity of a face in a surveillance image, the method comprising: (a) designating predetermined facial coordinate points capable of detecting a facial expression change in a face image of an approaching person photographed; (b) tracking a movement path of the designated facial expression coordinate points according to the facial expression change of the accessor in the facial image transmitted in real time; (c) determining whether the movement path of each of the facial expression coordinate points has the same direction when the facial expression changes; And (d) determining the approacher's face as a virtual face when the movement paths of each of the facial expression coordinate points have the same direction; Characterized in that it comprises a.

Surveillance, face authentication

Description

{METHOD FOR TRUTH OR FALSEHOOD JUDGEMENT OF MONITORING FACE IMAGE}

1 is an exemplary view of a system to which the present invention is applied.

2A to 2C are diagrams illustrating facial expression coordinate points in various facial expressions according to embodiments of the present invention.

3 is an overall flowchart illustrating a method of distinguishing the authenticity of a face in a surveillance image according to an exemplary embodiment of the present invention.

4 is a detailed flowchart illustrating a facial expression coordinate point tracking and result analysis step according to an exemplary embodiment of the present invention.

The present invention relates to a method for distinguishing the authenticity of a face in a surveillance image, and more particularly, in a surveillance system for determining whether a face is approached by detecting a face of a human body, whether the face is a real face image of a human body or a photograph or a manikin. The present invention relates to a method of distinguishing the authenticity of a face from a surveillance image that can directly determine whether the image is a virtual face image by using a captured image to increase the security of the surveillance system.

In general, face detection masks the presence or absence of a person's face in a given image for any input image, and if a face exists, finds each person's face in the image and displays its location. It can be used for a mug shot maching system, a search system using face information, and an object-oriented coding system.

Such a face detection system aims to identify an individual using physical characteristics unique to each person, such as fingerprint recognition and iris recognition. In particular, the fingerprint recognition technology, which is widely used recently, is expected to be commercialized enough that the fingerprint recognition devices are widely used in the entrances of luxury apartments or offices, and the demand for the fingerprint recognition devices is expected to increase further. Because of this, many people approach the place often has its drawbacks.

Moreover, with the increasing diversity of applications requiring user authentication, such as door locks in apartments or private homes, automobiles, ATMs, office doors, membership clubs, etc., the face recognition device is a conventional technology of fingerprint recognition device, IC / magnetic, etc. When combined with assistive recognition devices such as / RF card readers, various types of commercial products can be implemented depending on the security level, which requires active research.

Among the detailed techniques used in the face detection method, the difference image is a technique capable of tracking the position of a moving object and is a very effective technique. That is, the moving object is tracked by finding the position of the target from the difference image between the background and the object. When the absolute value of the difference image is obtained, the value is small in the case of the background image without the object, It will have a value. By creating a binarized image with the appropriate thresholds, you can effectively track the object by selecting this binarized region.

In addition, there is a template matching for detecting a face among techniques used in the face detection method.

Template matching, which is a conventional method, is a method of determining a position at the most similar part by comparing an image and a template with each other. This is one of the methods widely used for detecting a face area in the face recognition field, and is used to detect only the face area after capturing an image in which a person exists.

In this method, a template having a human face shape moves through the entire area of the image, compares the pixel values of the template with pixels of the image to obtain similarity, and then determines the area having the minimum value as the face region.

In the face detection, since the illumination changes from time to time and place, there is a method for smoothing the histogram to have a uniform distribution by redistributing the distribution of contrast values in the face region to reduce the effect on the illumination.

This method is used to minimize the influence of lighting in the face recognition field by making the distribution of contrast uniform.

Since the conventional face detection method described above is already known, a detailed description thereof will be omitted.

Here, the face monitoring system that determines whether the user is accessible by comparing the face image with the face biometric information of the registered accessor according to the detection of the face image and determining whether the object is classified as an accessor is special security. It is installed for the purpose of blocking public access to the gate or a specific system in a place where it is needed.

However, in recent years, accidents that attempt to access the face surveillance system using an accessor's photo or a precise manikin have frequently occurred, which is becoming a fatal problem.

In reality, there was no way to prevent the access of the accessor's photo or precisely manufactured manikin, but the infrared discrimination method that uses the heat distribution to break whether the accessor is a real person. However, a three-dimensional imaging technique is used to photograph three-dimensional information by using a feature that a real face is three-dimensional.

However, the above-described conventional infrared discrimination method or three-dimensional imaging technique to determine whether or not the approach of a real person is inefficient because the equipment must be equipped with a very expensive equipment in terms of cost in the manufacturing of equipment is inefficient and the equipment is also large and general purpose There are many challenges to being used as an enemy.

The present invention has been made to solve such a problem, the object of the present invention in the surveillance system to determine whether or not the approach of the face through the detection of the human body whether the real face image of the human body or a virtual face image of the manikin It is to provide a method of distinguishing the authenticity of the face from the surveillance video that can be directly identified through the photographing image, which can dramatically increase the security of the surveillance system at a small additional cost.

The method for distinguishing the authenticity of the face in the surveillance image of the present invention for achieving the above object comprises the steps of: (a) designating predetermined facial coordinate points that can detect a facial expression change in the face image of the approaching person; (b) tracking a movement path of the designated facial expression coordinate points according to the facial expression change of the accessor in the facial image transmitted in real time; (c) determining whether the movement paths combining the movement positions of each of the facial expression coordinate points have the same directionality when the facial expression changes; And (d) determining the approacher's face as a virtual face when the movement paths of each of the facial expression coordinate points have the same directionality; Characterized in that it comprises a.
Here, the movement paths of each of the facial expression coordinate points when the facial expression changes have the same directionality, which means a feature of the virtual face in determining the authenticity of the face of the surveillance image. When changing, they represent different movement paths without leaving the same movement path, whereas virtual faces such as face photographs and mannequins cannot change their expressions, so all facial coordinate points on the face have the same movement path. Since there is no moving path of facial expression coordinate points, it is an important criterion for determining the authenticity of the face.

Preferably, after the step (b), (e) comparing the moving path of each of the facial expression coordinate points with the predicted path which is a change in the movement path of the facial expression coordinate points that are generally present at the corresponding facial expression; And (f) determining the approacher's face as a virtual face when the movement path of each of the facial expression coordinate points is different from the prediction paths, which are changes in the movement paths of the facial expression coordinate points, which are generally displayed at the time of expression. step; It characterized in that it further comprises.

More preferably, the facial expression coordinate point is a specific coordinate on the face to be tracked for authenticity in the photographed face image, the start and end points of the eyebrows, the center point of the pupil, the left and right symmetrical points in each of the left and right eyes and It is characterized in that at least one or more of the upper and lower points, the symmetrical left and right points and the upper and lower points of the lips.

More preferably, before the step (b), it is determined whether there is a change of expression on the face of the accessor through the change in the movement path of the coordinates of the facial expression point, and if there is no expression change on the face of the accessor, Requesting a specific facial expression from the person; It characterized in that it further comprises.

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention;

1 is a block diagram showing a video surveillance system to which the present invention is applied.

Referring to FIG. 1, a video surveillance system to which the present invention is applied, the photographing unit 10, the main control unit 20, the image analyzing unit 30, the facial expression analyzing unit 40, the information storage unit 50, and a message The output unit 60, the security access object 70 is configured.

The photographing unit 10 captures an image of a person who wants to access a security access object 60 whose access is controlled by the video surveillance system at a place where the video surveillance system of the present invention is installed.

The image analyzer 30 stores an image of an accessor photographed by the photographing unit 10 and masks the presence or absence of a human face in the corresponding image with respect to the corresponding input image. It finds each person's face and displays its location and compares the biometric information of the registered accessor through template matching to determine whether the image of the accessor's face is the accessor's face. Since it is not a gist of the invention and is well known, a detailed description thereof will be omitted.

The facial expression analysis unit 40 is whether the face image of the accessor found through the image analysis unit 30 in the image photographed by the photographing unit 10 is a real face image of the human body or a photograph or a virtual face image of a manikin The facial expression analyzing unit 40 sets a plurality of facial expression points for determining the authenticity of the face in the face image of a person and moves the movement trajectory of the facial expression coordinate point. By comparing the relationship between the expected movement of the corresponding facial coordinate point according to the change of the facial expression of the real person and the actual movement of the facial expression coordinate point in the captured image, the authenticity of the accessor's face image in the captured image is compared. Will be determined.

The authenticity as used herein refers to determining whether the face image is a real human face or a virtual face (for example, a face photograph or a mannequin face).

The facial expression coordinate point is a specific coordinate on the face that is tracked to determine the authenticity of the photographed face image, the starting point and the end point of the eyebrow, the center point of the pupil, the left and right symmetric points in each of the left and right eyes, and It may be a top and bottom, symmetrical left and right and the top and bottom of the lips. An example of 18 facial expression coordinate points in such a face image is shown in various expressions in FIGS. 2A to 2C.

In fact, assuming that a person smiles at the expressionless face, the face of the person does not move only one part, but the eyes, nose, and mouth are related to each other according to a specific expression. Therefore, tracking their movements can distinguish whether a face in the captured image is a real face or not.

That is, after determining the coordinates of the portion of interest around the eyes and around the mouth of the image taken as described above, ie, the facial expression coordinate point, the movement trajectory of the facial expression coordinate point in accordance with the facial expression change of the face image is tracked and then applied. The authenticity of the face image is determined.

Here, the reason for excluding the nose from the facial coordinate point is that the facial expression change of a person's face actually occurs largely around the eyes and the mouth, and the nose almost does not change the coordinate according to the facial expression change.

In more detail, the authenticity determination process of the face image is described. Each coordinate in each of the facial expression coordinate points is determined to have a symmetrical structure, and thus has a characteristic of being reversed in the left and right areas when the facial expression changes. For example, in the case of a smile, the coordinates around the eyes and the coordinates around the mouth are symmetrical, and even if they are not symmetrical, they move in different directions when the facial expression changes. Therefore, by comparing them with each other, it is possible to determine whether they are real faces or not.

Of course, when there is no change in facial expression, the message output unit 60 may give a command such as to make a smile to forcibly induce facial expression change to determine whether or not the actual face is present.

In the system for estimating the movement path of the facial expression coordinate point, the virtual face, that is, the movement of the virtual face, that is, the face of the photo or mannequin, is tilted or tilted up and down as described above. Since it moves constantly in the direction, it is discriminated as a virtual face. Of course, if you rotate a photo or a mannequin's face in front of the camera, it may be judged that the facial expression coordinate points do not all have the same direction. Because it cannot move symmetrically, it is against the symmetry, so it is judged that it is not a real face.

Therefore, it is not just to determine whether or not it is a real face using the movement of facial coordinate points, but also to determine the actual face in consideration of the direction of movement of the facial coordinate points. Even if it is possible to determine whether the actual face.

The criterion for determining the authenticity of the photographed face image using the facial expression coordinate point is stored and used as facial expression prediction information in the information storage unit 50.

First, as a criterion for determining the authenticity of the photographing face image, it may be used whether or not the directionality is the same as the movement change of facial expression coordinate points. That is, as described above, when the facial expressions change, the corresponding facial coordinate points are moved, but if all the facial expression coordinates are moved in the same direction, the virtual face is determined, and when the symmetrical facial coordinate points are moved in different directions, the facial face is determined as the real face. do.

This means that if a person's face smiles like a smile, the facial coordinate points around the mouth of symmetry are moved in different directions as the facial expression changes, and the facial coordinate points around the symmetrical eyes are also different. It is true that For example, when smiling, it can be seen that the facial expression coordinate points on the left and right sides of the lips are further moved from side to side, respectively, to have different movement directions.

Next, as a criterion for determining the authenticity of the photographed face image, it may be used whether or not there is organicity in the movement change of the facial coordinate points. That is, when the facial expression changes, the corresponding facial coordinate points are moved, but because they have an organic relationship between the regions, the facial coordinates are moved to determine the actual face. In other words, organicity, which is a criterion for determining the authenticity of the photographed face image, is defined as movement path data corresponding to each of the facial expression points in each expression when the expression of the general public changes.

It is true that if a person's face makes a certain expression, such as a smile on a real person, the lips move together instead of only the eyes. In other words, when smiling, the width of the upper and lower facial coordinate points of the eyes become closer and the width of the left and right facial expression coordinate points of the lips gradually decrease so that the eye area and the lip area have an organic relationship with each other when the facial expression changes. Facial expression coordinate points are also moved in an organic relationship.

Next, as a criterion for determining the authenticity of the photographing face image, it may be used whether or not there is generality in the change of movement of the expression coordinate points. That is, when the facial expression changes, the corresponding facial coordinate points are moved, but when it is changed in the predictable path, the face is determined as a real face. In other words, generality, which is a criterion for determining the authenticity of the photographed face image, is defined as the movement path data of the expression coordinate point that appears when the general expression of the face.

It is true that facial coordinates are moved to predictable paths when the face of a real person smiles like a smile, and the authenticity of the face is compared with some predicted movement paths of these predictable facial coordinate points. It will determine whether or not. The similarity of the moving paths in comparison with the actual moving paths of the expected moving paths to the facial expression coordinates, that is, the similarity criteria for determining the authenticity, is calculated as the judgment distribution of the progression of the similar paths of the facial expression coordinates in the whole facial coordinates. Criteria for determining face authenticity based on the number and density of distributions can be applied in various ways depending on the specifications and security requirements of the system.

The authenticity discrimination criteria of the photographing face image using the above-described facial expression change points may be selectively applied to various systems.

The facial expression analyzer 40 performing the above function receives and stores the image photographed by the photographing unit 10 through the image storage unit 41, and is determined through the coordinate tracking unit 42. The movement path of the expression coordinate points are tracked, and the movement path coordinates of each expression coordinate point tracked through the tracking coordinate output unit 43 are transferred to the main control unit 20, and the main control unit 20 is provided. It is possible to determine the authenticity of the face by comparing the movement path coordinates of each of the facial expression coordinates with the authenticity determination criteria of the face image described above.

In this case, the information storage unit 50 has the data on the authenticity determination criteria of the face image as facial expression prediction information, and also has the biometric information of the person allowing access to the corresponding security access object 70. The image analyzer 30 determines whether the photographed person is an access permission.

As already mentioned, the message output unit 60 performs a function of instructing a person to be photographed to make a specific facial expression, and under the control of the main control unit 20, a text or an image through a voice or a display through a speaker. As a result, the forced facial expression command is executed for the accessor.

The method of distinguishing the authenticity of the face in the surveillance image of the present invention made through the system having the above-described configuration will now be described with reference to FIG. 3.

First, the information storage unit 50 stores biometric information about a face of a predetermined access granter with respect to the security access object 70 (S10), and the biometric information of the access granter is a face image of the accessor photographed thereafter. It is the basis for determining whether the accessor is an accessor through whether or not it is the same as.

In addition, the photographing unit 10 captures an image of an accessor who wants to access the security access object 70 and transmits the image to the main control unit 20 in real time (S20).

Accordingly, the image analyzing unit 30 recognizes and detects the face of the accessor based on the captured image of the photographing unit 10 (S30). Such a face recognition and detection technique is described in the present invention. Since it is not the gist of the present invention and is well known, a detailed description thereof will be omitted.

Next, the main controller 20 continuously transmits the image image of the detected face to the facial expression analyzer 40 according to time, and accordingly, the facial expression analyzer 40 displays the image image from the facial image image. By setting a specified facial coordinate point and tracking the movement path of each of the facial expression coordinate points set through the face image image continuously transmitted by time to report the movement path of the tracked facial expression coordinate points to the main control unit 20. (S40).

The main control unit 20 determines whether there is a facial expression change on the face of the accessor through the change of the movement path of the transferred facial expression coordinate points (S50), and when there is no facial expression change on the face of the accessor, The main control unit 20 requests the accessor to make a specific facial expression through the message output unit 60 (S51).

In this case, the message output unit 60 requests a specific facial expression to the accessor through voice, text, or video. Here, a specific expression may be a variety of expressions such as a smiling face (grin), angry face, grimace, crying expression.

On the other hand, as a result of the determination in step S50, if there is a facial expression change in the face of the accessor, the main control unit 20 is the facial expression coordinates according to the facial expression change in the facial image of the accessor through the facial expression analysis unit 40 The movement paths of the tracks are analyzed and the characteristics of the movement paths of the tracked facial expression coordinate points are compared with the expression prediction information of the information storage unit 50 (S60).

In this case, the facial expression prediction information is reference information for determining the authenticity of the photographing face image using the facial facial coordinate points, and as described above, the moving direction of the facial facial coordinate points, the organic nature of the movement change of the facial facial coordinate points, and the facial facial coordinate points The generality of movement change is based on that standard.

Herein, the step S60 related to the expression coordinate point tracking and the result analysis will be described in more detail.

First, the main controller 20 transmits the accessor face image in the surveillance image detected by the image analyzer 30 to the facial expression analyzer 40, and the image storage unit 41 according to time. Images of the face image to be transferred are stored (S61).

The coordinate tracking unit 42, which has received the image of the face image from the image storage unit 41, sets predetermined facial coordinate points as shown in FIG. 2 in the image of the face image (S62). According to the image of the face image transmitted according to the trace path that the corresponding coordinate coordinate points are moved (S63).

The coordinate tracking unit 42 transmits the information about the movement paths of the corresponding facial expression coordinate points to the main control unit 20 through the tracking coordinate output unit 43 (S64).

Accordingly, the main controller 20 determines the facial image authenticity of the face information of the information storage unit 50 based on the movement path information of the expression coordinate points according to the expression change in the accessor face image analyzed by the facial expression analyzer 40. The authenticity of the facial image is determined based on the movement direction of the facial expression coordinate points, the change organicity of the movement of the facial expression coordinate points, and the general variation of the movement of the facial expression coordinate points in comparison with the facial expression prediction information as a reference (S65).

Therefore, through the facial coordinate point tracking and analysis of the results, the system can directly determine whether the face image of the accessor is a real face image of the human body or a virtual face image of a manikin using only the face image of the approaching person. It becomes possible.

Meanwhile, the main controller 20 determines whether the face of the accessor is a real face through the tracking of the facial coordinate point and the analysis of the result of the above-described step S60 (S70). The procedure will be performed again by implementing.

In operation S70, when the face of the accessor is a real face, the main controller 20 may convert the detected face image into face biometric information previously registered as an access permitter of the information storage unit 50. In comparison, whether the corresponding accessor is a person having access to the security access object 70 is determined (S80), and if the accessor is allowed to access the security access object 70 of the accessor (S90). ).

Therefore, according to the detection of the face image, the face image is compared with the face biometric information of the pre-registered access person, and it is determined whether or not the target is classified as an access granter. The face authentication system installed in the face authentication system installed for the purpose of blocking the access of the gate or a specific system in the place where it is needed, the facial direction by the direction of movement of the facial expression coordinate points, the organic nature of the movement of the facial expression coordinate points, and the general movement of the facial expression coordinate points. Since the authenticity of the image can be determined, access attempts using the accessor's photo or sophisticated manikin can be blocked at the source.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains. It is not limited to the drawing.

As described above, the method for distinguishing the authenticity of the face from the surveillance image shown in the present invention is to set the facial expression coordinate points for the authenticity of the face image in the accessor's face image in the face authentication system and change the facial expression of the accessor. By tracking the movement paths of the corresponding facial coordinate points, the authenticity of the facial image can be directly determined by the direction of movement of the facial expression coordinate points, the organic nature of the movement of the facial expression coordinate points, and the general movement of the facial expression coordinate points. It is possible to increase the security of the system by blocking the access attempt by using the accessor's photograph or the elaborate mannequin's virtual face without using the infrared discrimination system or the 3D photographing system.

Claims (4)

(a) designating predetermined facial coordinate points capable of detecting a facial expression change in the face image of the approaching person photographed; (b) tracking a movement path of the designated facial expression coordinate points according to the facial expression change of the accessor in the facial image transmitted in real time; (c) determining whether the movement paths combining the movement positions of each of the facial expression coordinate points have the same directionality when the facial expression changes; And (d) determining the approacher's face as a virtual face when the movement paths of each of the facial expression coordinate points have the same directionality; Method for distinguishing the authenticity of the face in the surveillance image comprising a. The method of claim 1, After step (b), (e) comparing the moving paths of each of the facial expression coordinate points with a prediction path which is a movement path change value of facial expression coordinate points generally appearing at the corresponding facial expression; And (f) determining the approacher's face as a virtual face when the moving path of each of the facial expression coordinate points is different from the predicted paths, which are changes in the movement paths of the facial expression coordinate points that are generally present at a corresponding expression, by more than a predetermined range. ; Method for distinguishing the authenticity of the face in the surveillance image, characterized in that it further comprises. The method according to claim 1 or 2, The facial expression coordinate point is a specific coordinate on the face to be tracked for authenticity in the photographed face image, the start and end points of the eyebrows, the center point of the pupil, the left and right points and the top and bottom points symmetrical in each of the left and right eyes, Method for distinguishing the authenticity of the face in the surveillance image, characterized in that at least one or more of the left and right symmetrical points and the upper and lower points. The method of claim 3, Before step (b), Determining whether there is a facial expression change on the face of the accessor through whether the facial expression coordinate points are changed, and if there is no facial expression change on the face of the accessor, requesting a specific facial expression from the accessor; Method for distinguishing the authenticity of the face in the surveillance image, characterized in that it further comprises.

Images