KR101057718B1 - How to determine whether the acquired image can be identified by the user - Google Patents

Abstract

The present invention provides a method of determining whether an acquired image can be identified by a user so that the user can use a financial automation device only when the user's face is identifiable: obtaining a user face image including only a face region from the user image and Extract the image for each part of each face part, calculate the shape information for each part, compare the calculated template information for each part with pre-stored template data for each part, A method of calculating a similarity rate for each face part and determining whether the acquired image can be identified by the user based on the calculated similarity rate for each face part is provided.

Financial Automation Equipment, User Face Image, Template Data

Description

{METHOD FOR DETECTING APPLICABILITY TO USER IDENTIFICATION OF OBTAINED IMAGE}

The present invention relates to a method for determining whether a user can identify an acquired image, and more particularly, to accurately photograph each part of a user's body in a user image generated by photographing a user who wants to use a financial automation device. The present invention relates to a method of determining whether the acquired image can be identified by the user so that the financial automation device can be used only when the financial automation device is used.

Recently, ATM (Automated Teller Machine) including ATM automatic teller machine for performing unmanned automated financial tasks has been widely used. However, such financial automation devices use only a password composed of a plurality of combinations of numbers to identify legitimate users.

In this case, there is a fatal disadvantage that an unjust user cannot cope with financial incidents caused by the theft of password.

On the other hand, in recent years, in order to minimize the occurrence of such a financial accident, a method of making it easy to find an unjust user in the event of a future accident by taking a picture of a user performing a financial task in a financial automation device.

However, if an unjust user performs financial work with a face mask, eye patch, glasses, a hat, or the like, it is difficult to identify a suspect by an image of the unjust user.

Therefore, it is determined that the image of the user who performs the financial affairs is exposed enough to identify who the face is and is a suitable image for face identification, and if it is a suitable image, the financial affairs can be continued. If it is impossible to conceal, it is necessary to apply a method to restrict financial operations.

Meanwhile, as another method for preventing financial accidents, a method of identifying a user using a fingerprint, an iris image, a face image, and the like of a user in addition to a password has been studied, and has been gradually introduced into financial automation devices.

In the method of identifying a user by face recognition using a face image, a user image can be obtained without requiring a user to access a separate photographing apparatus or recognize a photographing moment, and user identification can be performed. It has been spotlighted as a user identification method.

However, in face recognition, a user's face image used for face recognition may be differently photographed by a photographing angle or ambient light of the user's face at the time of photographing, and thus the face recognition success rate may change. Therefore, for accurate face recognition, each part of the user's face must be accurately photographed. Otherwise, a method of inducing a user to re-photograph the face is required.

In addition, there may be a case in which an invalid user attempts to identify or authenticate a user by using an image photographing a legitimate user, so that there is a need to detect such a situation and make an appropriate response.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problem, and determines whether a face of a user who wants to use a financial automation device is photographed to be used for face recognition or is photographed to the extent that the user can be identified by the naked eye. The purpose is to provide a way to.

Another object of the present invention is to provide a method for determining whether a photographed user's face image is stolen by an illegal user.

According to an aspect of the present invention, there is provided a method for determining whether a user can identify an acquired image so that the user can use the financial automation device only when the user's face is identifiable. Photographing a user's body to generate a user image; Obtaining a user face image including only a face area from the generated user image; Extracting an image for each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; If the calculated similarity rate for each facial part is greater than the standard similarity rate for each part for the corresponding part, the user can use the financial automation device, and if the user is small, the user can use the financial automation device. Provided is a method for determining whether a user can identify an acquired image in a financial automation device.

In addition, the step of disabling the financial automation device may include: calculating a combined similarity rate by combining similarity rates for each of the face parts of a plurality of parts among the calculated similarity rates for each face part by a predetermined arithmetic formula; Making; Allowing the user to use the financial automation device when the combined similarity rate is greater than the reference combined similarity rate for the corresponding part, and when the combination similarity rate is small, the user cannot use the financial automation device. It is characterized by.

In addition, the step of making the financial automation device unusable may include: calculating a similarity rate for each face part with respect to all of the part-specific images extracted from the user face image; Calculating a total similarity rate by combining all of the calculated similarity rates for each facial part by a predetermined arithmetic formula; If the calculated total similarity rate is greater than a reference total similarity rate, allowing the user to use the financial automation device; and if the calculated total similarity rate is small, the user may not use the financial automation device. It is done.

In addition, the arithmetic expression combining the identifiable suitability for each part is characterized in that it can be arbitrarily set.

The method may further include setting a reference similarity rate for each site.

The method may further include extracting a background image excluding a user's body part from the generated user image; Extracting at least one line shape from the extracted background image; The method may further include disabling the financial automation device when the extracted linear form forms a closed figure surrounding the user face part of the user image.

The method may further include extracting a body outline of the user's body from the generated user image; The method may further include disabling the financial automation device if the user does not overlap the lower edge of the user image among the extracted body outlines.

In addition, the present invention provides a method for determining whether a user can identify the acquired image so that the user can use the financial automation device only when the user's face is identifiable: photographing a body of a user who intends to use the financial automation device. Generating a user image; Obtaining a user face image including only a face area from the generated user image; Extracting an image of each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; Calculating a combined similarity rate by combining the similarity rates for each of the face parts with respect to a plurality of predetermined areas among the calculated similarity rates for each face part by a predetermined arithmetic formula; Enabling the user to use the financial automation device when the calculated combined similarity rate is greater than the reference combined similarity rate for the corresponding part, and if the calculated combination similarity rate is small, the user cannot use the financial automation device. A method of determining whether a user can identify an acquired image is provided.

In addition, the present invention provides a method for determining whether a user can identify the acquired image so that the user can use the financial automation device only when the user's face is identifiable: photographing a body of a user who intends to use the financial automation device. Generating a user image; Obtaining a user face image including only a face area from the generated user image; Extracting an image of each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; Calculating a total similarity rate by combining all of the calculated similarity rates for each facial part by a predetermined arithmetic formula; If the calculated total similarity rate is greater than a reference total similarity rate, allowing the user to use the financial automation device; and if the calculated total similarity rate is smaller, the user cannot use the financial automation device. A method for determining whether a user can be identified is provided.

In addition, the present invention provides a method for determining whether a user can identify the acquired image so that the user can use the financial automation device only when the user's face is identifiable: photographing a body of a user who intends to use the financial automation device. Generating a user image; Extracting a background image excluding a user's body part from the generated user image to extract at least one line shape from the extracted background image, wherein the extracted line shape forms a closed figure surrounding the user face part of the user image. Or extracting a body outline of the user's body from the generated user image so that the lowest line among the extracted body outlines does not overlap with a lower edge of the user image, or the user's body shape from the generated user image. Extracting at least one line form from the extracted background image by extracting a background image from which the extracted line image is extracted; One case When determining that the photographed image is compromised by the user video, and if not true for any of the above, if the user determines that the image generated by photographing a normal user; When the user image is determined to be a normal image, acquiring a user face image including only a face area from the user image; Extracting an image of each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; Calculating a combined similarity rate by combining the similarity rates for each of the face parts with respect to a plurality of predetermined areas among the calculated similarity rates for each face part by a predetermined arithmetic formula; Enabling the user to use the financial automation device when the calculated combined similarity rate is greater than the reference combined similarity rate for the corresponding part, and if the calculated combination similarity rate is small, the user cannot use the financial automation device. A method of determining whether a user can identify an acquired image is provided.

According to the method for determining whether a user can identify the acquired image according to the present invention having the above-described configuration, it is possible to determine whether a user who intends to use the financial automation device is covering part of the face unfairly. Therefore, it is possible to determine a case in which an unjust user wants to execute a financial task while covering the face so that the face cannot be identified, thereby reducing the possibility of an unfair transaction.

In addition, when a part of the face part of a legitimate user is inevitably covered by a disease or an ornament, etc., when some part of the face part is indistinguishable, it can respond flexibly by applying various criteria.

EMBODIMENT OF THE INVENTION Hereinafter, this invention is demonstrated in detail with reference to attached drawing.

The apparatus 100 for determining a user identifiable according to an embodiment of the present invention may capture a face of a user by capturing a face of a user or a system that performs user identification using face recognition, such as an ATM or a digital door lock. Equipped with a system that allows the user to continue using the device only if his or her identity is clearly identifiable, used to determine if the user's face in the image of the user's image has been captured to be used for face recognition or user identification. do. If it is determined that the image photographed by the user identifiable determining apparatus 100 is properly captured to be user identifiable and applied to face recognition, a separate face recognition means performs user identification by face recognition. If it is determined that the captured image was captured in a state that is hidden so that the user cannot be identified, or that the image was not captured for face recognition, additional guidance such as blocking access to the user or asking the user to retake the image may be performed. Can be.

In addition, the user identifiable determination apparatus 100 according to an embodiment of the present invention is a procedure that may be added before or after determining whether the user identifiability or face recognition is possible as described above. And determining whether the captured image is captured from an image stolen by an illegal user. If it is determined that the image is stolen, the user is asked to retake the picture or output a separate security alert and block access to the system to be used. On the other hand, if it is determined that the captured image is a normal image without theft, it is possible to continue using the system to be used.

As such, the user identifiable determining device 100 according to the present invention is coupled to any system including a financial automation device, and operates under the control of the system. In other words, after receiving the captured image from the system for user identification, it is determined whether the image is a suitable image and / or a stolen image, and only the determined result is output. Of course, a separate photographing means may be provided and the user may be photographed directly as needed.

1 is a block diagram showing the configuration of a user identifiable determining apparatus for executing a method for determining a user identifiability of an acquired image according to an embodiment of the present invention. According to FIG. 1, the user identifiable determining apparatus 100 includes an image extracting unit 10, a similarity rate calculating unit 20, a template storage unit 25, an identifiable determining unit 30, and an illegal plate for each part. It includes the government 35.

In the following description, the identifiable determination device according to the present invention will be described as a case of operating as a part of the ATM.

The part-specific image extraction unit 10 receives a user image generated by photographing a user requesting the use of the device from a financial automation device to which a separate photographing means (not shown) or a user identifiable determination device is applied, A user face image corresponding to the face region of the user is obtained by analyzing the provided user image, and an image for each part including each part constituting the face is extracted from the obtained user face image. The extracted region-specific image is provided to the similarity rate calculator 20 to be described later.

The similarity rate calculation unit 20 calculates shape information (for example, numerically representing a feature point) of a face part included in a part-specific image provided corresponding to each part of the user's face. The similarity ratio between the calculated shape information and the template data for each part previously stored in the template storage unit 25 is calculated to calculate the similarity rate for each face part.

In the template storage unit 25, template data divided for each part of the face is maintained as template data for each part.

The identifiable determining unit 30 is a combination of the calculated similarity rate for each face part alone, or a combination of similarity rates for each face part for a plurality of parts, or a combination of similarity rates for each part for all parts. After that, if the calculated value is larger than the standard fit ratio, it is determined that the provided user image is a suitable image that can be used for user identification or face authentication so that the financial automation device can be continuously used. On the other hand, in other cases, it is determined that the user's face is blocked or the facial expression is changed so that the user identification or face recognition is impossible and the use of the financial automation device is blocked.

Here, each part of the face image refers to all body parts including eyebrows, eyes, nose, mouth, ears, jawline, cheekbone contour, wrinkles of a specific part, and the like. Moreover, eyebrows, eyes, ears, etc. are analyzed by distinguishing between left and right sides.

The section-specific image refers to a reconstruction of the user's face image into an image including only one of the sections. That is, the position corresponding to each part of the user's face image is roughly designated first, and the shape of the corresponding part is precisely identified and extracted as the image for each part by analyzing the color, saturation, and contrast in the designated area.

The form-specific information for each region refers to digital information obtained by extracting feature points in order to use face parts included in the region-specific image for face recognition, analyzing the feature points by a vector method or a bitmap method, and converting the feature points into numerical information. The method of quantifying the feature point information can be applied to a variety of commercially available techniques.

The site-specific template data refers to data about an image to be used as a reference for detecting each site within a digital image. The template data for each region extracts an image of the corresponding region from a variety of face images into an image having an arbitrary M × N size, and generates information indicating a reference form of the corresponding region by averaging the shape of each facial region of the extracted image. do. Alternatively, a feature point may be extracted from the corresponding part image, and the feature point may be stored as template data for each part.

The user identifiable determination device 100 according to an embodiment of the present invention may further include the theft determining unit 35.

The theft determining unit 35 determines whether the user image is captured from the stolen image by using the background image excluding the user face and the user's body part from the provided user image. That is, a line shape including a straight line shape is extracted from the background image, and the extracted line shape is analyzed to determine whether the image is stolen.

Such an identifiable determination unit 30 and the theft determination unit 35 output the determination result to the control means (not shown) of the financial automation device, and the control means performs a process corresponding to the provided determination result. . Here, the determination of whether the user can be identified and whether the theft image is performed may be performed simultaneously or continuously in a predetermined order, which may be freely set by the designer.

Next, with reference to FIGS. 2 to 8, a method of determining whether an acquired image is photographed so as to be able to discern a user's face using the user identifiable determination apparatus having the above-described configuration will be described in detail.

FIG. 2 is a diagram illustrating an overall flow of a method of determining whether an acquired image can be identified by a user according to an embodiment of the present invention. First, the user identifiable determination apparatus 100 is provided with a user image photographing the user (S10). In operation S20, a user face image including only a user face part may be acquired from the provided user image. When the user's face image is obtained, an image for each part is extracted by analyzing an image for each part forming a face from the user's face image (S30).

Subsequently, shape information for each region is calculated by specifying and quantifying feature points of the included facial regions or generating and quantifying a vector representing the shape of the facial region from the extracted region-specific image for each region (S40). When the shape information for each part is calculated, the degree of similarity indicating the degree of coincidence for each part is calculated by comparing the information with the shape information for each part (template data for each part) which is a reference, and based on the similarity for each face part. The similarity rate is calculated (S50). The similarity rate for each face part is that each part of the face is photographed at the correct angle with respect to the camera, under the correct illumination, and at the correct resolution, so that the face image included in the generated user image is used for face recognition for user identification. Or a scale indicating whether or not the user can be identified by the extracted user face image.

On the other hand, when the similarity rate for each facial part is calculated for each part as described above, the size of the calculated similarity rate for each facial part is compared with the reference similarity rate for each part, which is a preset standard similarity rate for each corresponding part ( S60). At this time, if it is determined that the similarity rate for every face part is larger than the corresponding similarity rate for each part, it is determined that the face is clearly photographed to the extent that the user can be identified using the photographed user image. If the financial automation device can continue to be used (S70), and at least one of the similarity rates for each facial part is smaller than the corresponding similarity rate for each part, an unfair user obscures a part of the facial part to the photographing means. Approach as determined to use the financial automation device to block the use of the financial automation device (S80).

In addition, the step of S80 may be caused by the financial automation device not properly photographing the user's face, so that a predetermined guide signal is displayed to the user to induce re-shooting, or the value of the reference similarity rate for each part of the corresponding part is determined. It may be reduced to recompare or proceed to the relaxed fit image determination step (A). Step A corresponds to FIG. 4.

As an example of such a situation, if the user has been photographed by a photographing means while covering the nose and mouth with a mask or the like while the reference similarity rate for each part is set to 70%, the face parts such as eyebrows, eyes, ears, etc. Although it can be used for face recognition, since the nose and mouth cannot be photographed, each identifiable similarity rate is 0%, thus determining that it is not available for financial automation equipment.

On the other hand, there may be a case where the similarity rate is low because the surrounding environment where the user is photographed is dark or the reflected light is generated by the light or the user wears a hat or the like so that a specific face part of the user is not accurately photographed (eg, For example, the similarity rate of each part of the face for the left eyebrow may be calculated as 70%). In this case, the relative similarity rate of each part of the eyebrow part may be reduced to about 60% to be flexibly responded to this situation. Make sure

Here, it is preferable that the reference similarity rate for each site is set as a lower limit to a level that can identify the user by identifying each face area at least by visual observation.

FIG. 3 is a flowchart for explaining a procedure for calculating a similarity rate for each facial part in FIG. 2. First, when shape information for each of at least one face part is calculated from the user face image (S14; corresponding to S40 of FIG. 2), the template storage unit stores the template data for each part that is preset and maintained in correspondence to the corresponding part. Acquiring by reading from (25) (S15), and comparing the acquired template data for each part and the calculated form information for each part (S16) to calculate the similarity, based on the similarity rate of similarity for each face part for the corresponding part It is calculated (S17).

4 is a flowchart for describing a procedure of determining a suitable image by combining similarity rates of facial parts for a plurality of parts. The procedure of FIG. 4 may be used as a method of applying relaxed criteria when it is determined in the procedure of FIG. 2 that it is not available for a financial automation device.

First, a similarity rate for each facial part calculated for each part constituting the face is obtained (S25), and the similarity rate for each facial part calculated for two or more parts is combined by a specific arithmetic formula. The similarity rate is calculated (S26).

Here, the arithmetic formula for calculating the combined similarity rate can be arbitrarily set. That is, the combined similarity rate may be calculated by a simple sum or by adding weights to each similarity rate, and calculating the combined similarity rate. The combined similarity rate may be calculated through various arithmetic operations or more complex arithmetic expressions. It may be. The calculated combined similarity rate may further include information indicating which facial parts have similar similarity rates.

The calculated combination similarity rate is compared with the reference combination similarity rate for each part used in the combination (S27), and when it is determined that the calculated combination similarity rate is greater than the reference combination similarity rate, the user face image of the user image is displayed. If it is determined that the image is suitable to be identified, it is determined that the financial automation device is available (S28). If it is determined to be small, the image is determined to be unsuitable image that cannot be identified by the user (S29).

Further, if it is determined that the image is not suitable, the user can be provided with a predetermined guide, or can proceed to the further relaxed suitable image determination step (B). The relaxed step B corresponds to FIG. 5.

For example, in this case, the combined similarity rate is calculated by adding and averaging the similarity rate for each face part of the left eyebrow, the right eyebrow, the left eye and the right eye in the user's face image, and the calculated combined similarity rates are both If the eyebrow and both eyes are larger than the preset reference combination similarity rate, it is determined that the user's identity can be identified by the naked eye or that the user's face image can be used for face recognition.

That is, when the user wears sunglasses or the like and both eyes are almost covered, the similarity rate of the left eyebrow is 80%, the similarity rate of the right eyebrow is 90%, and the similarity rate of the left eye and the right eye is 10%. have. At this time, the combined similarity rate calculated by the simple sum and average is 47.5%. On the other hand, if the reference combination similarity rate for both eyebrows and eyes is set to 60%, it is determined that the user face image generated by photographing the user is an inappropriate image.

On the other hand, when the user puts an eye patch on one eye, the similarity rate of the left eyebrow is 80%, the similarity rate of the right eyebrow is 90%, the similarity rate of the left eye is 80%, and the similarity rate of the right eye is 0%, and the combined similarity rate is 62.5%, which can be determined as a suitable image.

By applying the combination similarity rate as described above, the user identification can be performed even if a part of the face part of the user is inevitably covered by a disease, an ornament, or a surrounding environment.

FIG. 5 is a flowchart for explaining a procedure of determining a suitable image after combining the similarity rate for each facial region for all facial regions. The procedure of FIG. 5 may be used as a method of applying a relaxed criterion when it is determined in the procedure of FIG. 4 that the image is not available for the financial automation device. It may also be applied as step A of FIG.

First, the similarity rate for each face part calculated for all parts constituting the face (the part used for face recognition) is obtained (S35), and the similarity rate for each and every face part are all combined by a predetermined arithmetic formula. The total similarity rate is calculated (S36). The calculated total similarity rate is compared with the reference total similarity rate (S37). If the total similarity rate is greater than the reference total similarity rate, the user's face image is determined as a suitable image (S38). It determines (S39). When the inappropriate image is determined as described above, a predetermined guide is output to notify that the financial automation device cannot be used because the user face image generated by photographing the user cannot be used for user identification.

By using the total similarity rate for the entire region as described above, a flexible response to the occlusion of the user's face is possible, similar to using the combination similarity rate described above.

According to the method of determining suitability for face recognition of the acquired image by the user-identifiable determination apparatus 100 including the image extraction unit 10 and the similarity rate calculation unit 20 for each part as described above, the user identification For this purpose, it is possible to determine whether a part of the user's face photographed by the photographing means is clearly and clearly captured at an accurate angle such that a part or the whole of the face can be recognized. In addition, it is also possible to determine whether part or all of the user's face is covered by an illegal purpose.

Therefore, when a user who wants to perform financial affairs in a financial automation device, etc. is covering his face unjustly, he or she can notify the administrator to confirm whether the user is using illegally by limiting the financial affairs smoothly or by reporting to a security center. And the like can be processed.

On the contrary, even when a part of the face is hidden, the user can be smoothly identified by using the remaining normally photographed part.

FIG. 6 illustrates whether or not the user image is stolen by the theft determining unit 35 in the method of determining whether the acquired image is identifiable using the user identifiable determination apparatus 100 according to an embodiment of the present invention. It is a flowchart for explaining the method of determining. Here, a method for determining a case in which a user identification is attempted using display means in which a user's face is displayed, in particular, using a photograph in which the user's face is taken, is disclosed.

First, referring to FIG. 6, the theft determining unit 35 extracts a background image of a background excluding a face and a body part of a user from a user image photographed by a photographing means (S51) and from the extracted background image. A linear form including a straight line and / or artificially formed curves is extracted (S52). Preferably, the straight form is extracted. The extracted line shape is analyzed to determine whether the user image is a stolen image or a normal image generated by directly photographing an actual user (S53).

7 and 8 are flowcharts illustrating a detailed method for determining whether the image is stolen in FIG. 6.

First, in FIG. 7, the theft determining unit 35 determines whether the linear form extracted from the background image forms a closed figure surrounding the user's face (S63, S64). In this case, when there is a closed figure surrounding the face of the user, the outer edge of the output, such as an ID photo including the face of the user and the body part of the user, or the edge of the display means on which the user's face is displayed, It means a state in which the image is captured to be seen in the image capturing area (see FIG. 10).

When a closed figure surrounding the face of the user is found in this manner, it is determined that the user's image is stolen by an unjust user (S65), and the use of the financial automation device can be restricted (S66).

Next, in FIG. 8, the theft determining unit 35 extracts a body outline representing the outline of the user's body, which is the inner boundary of the background image (S72), and the bottom line of the extracted body outline is the user's image. It is examined whether or not to overlap with the bottom border (S73, S74). As a result of the examination, when it is confirmed that the bottom line of the extracted body outline does not overlap with the bottom border of the captured user image, it is determined that theft is attempted by the image of the body of the user. (S75). On the other hand, if the bottom line of the extracted body outline overlaps the bottom edge of the captured user image, it is determined that the user is normally photographed (S76).

In this case, as illustrated in FIG. 11, when an unjust user steals a photograph of a legitimate user's face and / or a body part and approaches the photographing means, a lower edge of the stolen photograph is used. Is a case where the image is viewed in the photographing area of the photographing means. For example, at least part of the user's chest is not shown because the ID photo is shown as a straight line in the outline of the body. If the straight line does not coincide with the lower edge of the user's image, the image is clearly displayed. It may be stolen.

The above-described procedure for determining image theft is described with reference to FIGS. 9 to 11.

First, FIG. 9 illustrates a user image normally taken by a camera in a state in which a legitimate user is exposed to all faces. There is no closed figure surrounding the user's face, and the bottom line of the body outline coincides with the bottom edge of the user image.

10 illustrates a state in which an invalid user attempts to identify a user by using a photograph taken by a valid user. Here, a rectangular linear form (part represented by a red square) has been extracted from a portion of the user image except for a facial portion (part represented by a green circle), and the extracted linear form surrounds the facial image. This state is determined to be the stolen image by the procedure of FIG.

Next, FIG. 11 is a state in which an invalid user attempts an illegal access to a picture of a legitimate user. Here, the body outline of the lower part of the chest of the user's body space is spaced apart without overlapping the bottom edge of the user image. Indicates. This state is determined to be the stolen image by the procedure of FIG.

Meanwhile, another method of determining the case where the stolen image is used will be described with reference to FIGS. 12A and 12B.

First, the theft determining unit 35 extracts and maintains a plurality of line shapes from the surrounding image generated by photographing the front in a state where there is no user at the current position of the financial automation device. Then, when the user approaches to use the financial automation device, if the line shape of the background image extracted from the user image does not match the line shape from the surrounding image previously held, it is determined that the stolen image is applied. For example, you can block subsequent requests.

FIG. 12A illustrates an example of the surrounding image and the extracted line form, and FIG. 12B illustrates an example of the user image and the extracted line form.

In the above-described method for determining whether the user can identify the acquired image according to the present invention, first, the similarity rate for each face part is determined for each part for each part, and a suitable image is determined by comparing with the reference similarity rate of the corresponding part, and the nonconformity is determined. When it is determined that the image is determined, the combination similarity rate or the total similarity rate is sequentially applied.

However, a configuration may be made such that any one of the similarity rate, the combined similarity rate, or the total similarity rate for each specific face part is completed to determine whether the user can be identified.

1 is a block diagram illustrating a configuration of a user identifiable determining apparatus for executing a method of determining a user identifiability of an acquired image according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an overall flow of a method of determining whether an acquired image can be identified by a user according to an embodiment of the present invention.

FIG. 3 is a flowchart for explaining a procedure for calculating a similarity rate for each facial part in FIG. 2.

FIG. 4 is a flowchart illustrating a procedure of determining a suitable image by combining similarity rates of facial parts of a plurality of parts.

FIG. 5 is a flowchart for explaining a procedure of determining a suitable image after combining all similarity rates for each facial part for all facial parts.

6 is a flowchart illustrating a method of determining whether the user image is stolen.

7 and 8 are flowcharts illustrating a detailed method for determining whether the image is stolen in FIG. 6.

9 shows a user image by a legitimate user.

10 and 11 illustrate a state in which an unjust user uses a picture taken by a right user.

12 (a) is a diagram showing an example of the surrounding image and the extracted line shape, Figure 12 (b) is a diagram showing an example of the user image and the extracted line form.

Claims (10)

delete A method of determining whether an acquired image is user identifiable so that a user can use a financial automation device only when the user's face is identifiable: Generating a user image by photographing a body of a user who intends to use the financial automation device; Obtaining a user face image including only a face area from the generated user image; Extracting an image for each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; If all of the calculated similarity rate for each face part is larger than the standard similarity rate for each part for the corresponding part, the user can use the financial automation device, and among the calculated similarity rates for each face part. If any one is smaller than the site-specific similarity rate for each of the corresponding sites, the combination similarity is obtained by combining similarity rates for each of the face parts for a plurality of sites among the calculated similarity rates for each face part by a predetermined arithmetic formula. Calculating a ratio, and allowing the user to use the financial automation device when the combination similarity rate is greater than a preset reference combination similarity rate for the plurality of parts, and the combination similarity rate is greater than the reference combination similarity rate. In small cases, the financial automation device may not be available to the user. Determining whether the acquired image of the acquired image includes a step. A method of determining whether an acquired image is user identifiable so that a user can use a financial automation device only when the user's face is identifiable: Generating a user image by photographing a body of a user who intends to use the financial automation device; Obtaining a user face image including only a face area from the generated user image; Extracting an image for each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; If all of the calculated similarity rate for each face part is larger than the standard similarity rate for each part for the corresponding part, the user can use the financial automation device, and among the calculated similarity rates for each face part. When any one is smaller than the site-specific similarity rate for the corresponding site, all of the calculated similarity rates for each face part are combined by a predetermined arithmetic formula to calculate a total similarity rate, and the calculated total similarity rate. If the rate is greater than the preset reference total similarity rate, the user can use the financial automation device. If the total similarity rate is less than the reference total similarity rate, the user cannot use the financial automation device. The method of claim 1, further comprising the step of identifying the user identification of the acquired image. The method according to claim 2 or 3, And the arithmetic expression combining the similarity rate for each face part can be arbitrarily set. The method according to claim 2 or 3, And determining the reference similarity rate for each part. A method of determining whether an acquired image is user identifiable so that a user can use a financial automation device only when the user's face is identifiable: Generating a user image by photographing a body of a user who intends to use the financial automation device; Obtaining a user face image including only a face area from the generated user image; Extracting an image for each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; If any one of the calculated similarity rates for each face part is smaller than the standard similarity rate for each part for the corresponding part, the user cannot use the financial automation device, and the similarity rate for each face part is calculated. When all of the values are larger than the reference similarity rate for each part of the corresponding part, a background image excluding a user body part is extracted from the generated user image, and at least one line shape is extracted from the extracted background image. If the extracted line forms a closed figure surrounding the user's face in the user image, the user cannot use the financial automation device, and if the closed figure is not found, the user may use the financial automated device. The user of the acquired image further comprising the step of enabling Determination by way possible. A method of determining whether an acquired image is user identifiable so that a user can use a financial automation device only when the user's face is identifiable: Generating a user image by photographing a body of a user who intends to use the financial automation device; Obtaining a user face image including only a face area from the generated user image; Extracting an image for each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; If any one of the calculated similarity rates for each face part is smaller than the standard similarity rate for each part for the corresponding part, the user cannot use the financial automation device, and the similarity rate for each face part is calculated. If all of the values are larger than the reference similarity rate for each part of the corresponding part, the body outline of the user's body is extracted from the generated user image, and the body outline of the extracted body outline is the bottom of the user image. The method of determining whether the user can identify the acquired image further comprises the step of disallowing the user to use the financial automation device when not overlapping with an edge, and allowing the user to use the financial automation device when overlapping each other. . delete delete A method of determining whether an acquired image is user identifiable so that a user can use a financial automation device only when the user's face is identifiable: Generating a user image by photographing a body of a user who intends to use the financial automation device; Extracting a background image excluding a user's body part from the generated user image to extract at least one line shape from the extracted background image, and the extracted line shape forms a closed figure surrounding the user's face part of the user image. Or extracting a body outline of the user's body from the generated user image, and the lowest body outline among the extracted body outlines does not overlap with a lower edge of the user image, or from the generated user image. After extracting the background image from which the shape has been removed and extracting at least one line shape from the extracted background image, the extracted line shape does not match the reference line shape extracted from the surrounding image photographed without the user at the current position. At least one Determining that the user's image has been stolen if the image is stolen, and determining that the user's image is generated by photographing a normal user if none of the above cases is true; When it is determined that the user image is taken from the stolen image, the user cannot use the financial automation device, and when it is determined that the user image is generated by photographing a normal user, only the face area is included from the user image. Obtaining a user face image; Extracting an image of each part of each face from the obtained user face image; Calculating shape information for each part digitized from the extracted images for each part; Calculating similarity rate for each face part by comparing the calculated shape information for each part and template data for each part previously stored for the corresponding part; Calculating a combined similarity rate by combining the similarity rates for each of the face parts with respect to a plurality of predetermined areas among the calculated similarity rates for each face part by a predetermined arithmetic formula; When the calculated combination similarity rate is greater than the reference combination similarity rate preset for the plurality of parts, the user can use the financial automation device, and the calculated combination similarity rate is smaller than the reference combination similarity rate. And determining that the user cannot use the financial automation device.

Images