Facial recognition-based identity authentication device

Abstract

According to the present invention, a facial recognition-based identity authentication device comprises: a storage unit for storing a plurality of pieces of feature information for each face area of a plurality of people; a communication unit for receiving feature information including a plurality of features of the face area generated by an authentication terminal from a captured face image of a person who requests identity authentication; an authentication unit for checking whether there is feature information having a similarity, which exceeds a preset threshold, with the feature information received via the communication unit from the storage unit and performing authentication; a classification unit for classifying, into group units according to the authentication result from the authentication unit, the feature information received from the authentication terminal and pre-stored target feature information exceeding a similarity threshold; and an adjustment unit for identifying the number of pieces of feature information of each group to adjust the number of pieces of feature information in each group.

Description

facial recognition-based identity authentication device

[0001] The present invention relates to an identity authentication device based on facial recognition.

[0002]

[0003] Identity authentication technology has evolved from visual verification of individuals and ID / password methods to a method that uses sensors to verify that each individual's unique biometric information matches, and biometric-based identity authentication technology is being used in various fields.

[0004] Biometric information used for identity verification has been primarily fingerprints or irises, but the authentication process has been somewhat cumbersome as the person requesting authentication must take additional actions for authentication, such as directly touching a sensor with a finger or bringing their eyes close to it.

[0005] Recently, with the development of image sensors and image processing technologies, facial recognition technology using cameras has been adopted as a new identity authentication technology, and such identity authentication technology using facial recognition has been disclosed in Republic of Korea Patent Publication No. 10-2024-0022161 (Application date: 2022.08.11., Publication date: 2024.02.20., hereinafter referred to as “prior art”).

[0006] However, in the prior art, when the result of identity authentication falls below the authentication standard value, additional feature information is requested from the user in addition to the user's facial recognition-based feature information used in the existing authentication, and identity authentication is performed using the accumulated and stored feature information. Whenever identity authentication for multiple users fails, the feature information for each user is accumulated and stored in the database more than necessary, and since identity authentication is performed through a comparison of the similarity between feature information, the identity authentication processing cost increases according to the number of accumulated and stored feature information, making it difficult to perform quick identity authentication.

[0007]

[0008] The present invention aims to provide a facial recognition-based identity authentication device that can minimize the time required for identity authentication while improving the accuracy of identity authentication in order to solve the above-described problems.

[0009]

[0010] According to one embodiment of the present invention, a facial recognition-based identity authentication device comprises: a storage unit storing a plurality of feature information extracted from each facial region of a plurality of persons; a communication unit communicating with an authentication terminal and receiving feature information including a plurality of features of a facial region generated from a photographed facial image of a person requesting identity authentication from the authentication terminal; an authentication unit confirming whether, among the plurality of feature information stored in the storage unit, there is feature information whose similarity with feature information received through the communication unit exceeds a preset threshold value, and transmitting an identity authentication result to the authentication terminal through the communication unit based on the confirmation result; a classification unit grouping and classifying, by person unit, feature information received through the communication unit and feature information whose similarity exceeds a threshold value stored in the storage unit only when the authentication unit confirms that the similarity exceeds the preset threshold value; And a control unit that checks the number of characteristic information of each group classified and stored in the storage unit by the classification unit and adjusts the number of characteristic information in each group depending on whether the number of characteristic information in each group exceeds a preset standard number; wherein the authentication unit checks whether the similarity between the characteristic information received through the communication unit and the plurality of characteristic information classified into groups by the classification unit among the plurality of characteristic information stored in the storage unit exceeds a threshold value, and the control unit compares the similarity between the characteristic information in each group only when the number of characteristic information in each group classified by the classification unit exceeds a preset standard number, and adjusts the number of characteristic information in each group by excluding the characteristic information with the lowest similarity between the characteristic information from the corresponding group.

[0011] Here, the above feature information is characterized as information extracted in the form of a vector value including features including eyes, nose, mouth, and facial outline in the facial area.

[0012] In addition, the control unit is characterized in that it checks the group classification time of each characteristic information by the classification unit for a group in which the number of characteristic information is greater than or equal to a preset minimum standard number in a predetermined periodic unit, and adjusts the number of characteristic information within each group by excluding the characteristic information with the oldest classification time from the group.

[0013] At this time, the control unit is characterized in that it deletes the characteristic information excluded from the group from the storage unit after a preset period of time.

[0014] On the other hand, the control unit is characterized in that, when the similarity between the characteristic information received through the communication unit and the characteristic information exceeding a threshold is confirmed through the authentication unit, the characteristic information excluded from the group to which the characteristic information belongs before the confirmation time is deleted from the storage unit.

[0015] Here, the authentication unit is characterized in that it additionally checks whether the similarity between at least one or more characteristic information excluded from the group by the control unit and the characteristic information received through the communication unit exceeds the threshold value only when the similarity between all the characteristic information classified and stored in the storage unit in groups and the characteristic information received through the communication unit is below the threshold value.

[0016] And, the authentication unit is characterized in that it identifies a group classified with the feature information having the highest similarity among a plurality of feature information having a similarity less than a threshold value with the feature information received through the communication unit, and it identifies whether the similarity between at least one feature information excluded from the group by the control unit and the feature information received through the communication unit exceeds the threshold value.

[0017] In addition, the classification unit is characterized in that, if it is confirmed that among at least one characteristic information excluded from the group through the authentication unit, the similarity between the characteristic information received through the communication unit and the characteristic information exceeds a threshold value, the classification unit reclassifies the characteristic information into an existing group.

[0018]

[0019] The facial recognition-based identity authentication device according to the present invention maintains a certain number of feature information, which is the subject of verification of whether the similarity threshold with the feature information of the person requesting identity authentication received from the authentication terminal, in a state in which the feature information is updated for each person, thereby preventing an unnecessary increase in the time and resources required for identity authentication confirmation, and since the feature information on the relatively recent face of the person subject to identity authentication is maintained, there is an effect of enabling rapid and highly accurate face recognition-based identity authentication.

[0020] In addition, the facial recognition-based identity authentication device according to the present invention does not immediately delete from the storage area, among the characteristic information of each person unit, the characteristic information that is oldest or has the lowest mutual similarity, but retains it until the time of identity authentication of the person in question, and then additionally utilizes it according to the result of the identity authentication of the person in question, thereby ensuring a certain level of identity authentication even when the face of the person in question changes.

[0021]

[0022] FIG. 1 illustrates an identity authentication device based on facial recognition according to a preferred embodiment of the present invention.

[0023]

[0024] Hereinafter, a preferred embodiment of a facial recognition-based identity authentication device of the present invention will be described in detail with reference to the attached drawings.

[0025] The same reference numerals in each drawing represent the same components. In addition, specific structural and functional descriptions of embodiments of the present invention are merely illustrative for the purpose of explaining embodiments according to the present invention, and unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by those of ordinary skill in the art to which the present invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning they have in the context of the relevant technology, and are preferably not interpreted in an ideal or excessively formal sense unless explicitly defined herein.

[0026] Before describing a facial recognition-based identity authentication device according to a preferred embodiment of the present invention with reference to the attached drawings, the identity authentication device of the present invention is configured to receive feature information from an authentication terminal in which a plurality of features for eyes, noses, mouths, and facial contours from facial images of a plurality of people are registered and stored as feature information in the form of vector values, and in which a camera is installed and connected to a wired / wireless communication network for communication, and an application is installed to generate feature information from a facial image of a person requesting identity authentication captured by the camera, and perform identity authentication of the person requesting identity authentication based on the degree of similarity between the received feature information and the feature information stored internally. At this time, the identity authentication device may be provided in the form of a computing device in the form of a server, and the configuration of the identity authentication device and the operation of each component to be described below should also be understood as being an abstract configuration and linked operation in which hardware and software are linked.

[0027]

[0028] FIG. 1 illustrates an identity authentication device based on facial recognition according to a preferred embodiment of the present invention.

[0029] Referring to FIG. 1, a facial recognition-based identity authentication device (100) according to a preferred embodiment of the present invention may include a storage unit (110), a communication unit (120), an authentication unit (130), a classification unit (140), and a control unit (150), in which a plurality of feature information extracted from the facial regions of a plurality of persons are classified and stored in groups according to each person, and communicates with an authentication terminal (200) to receive feature information including a plurality of features of a facial region generated from a photographed facial image of a person (O) requesting identity authentication, and performs identity authentication by comparing the feature information of the person (O) requesting identity authentication with previously stored feature information.

[0030] The storage unit (110) can store multiple feature information extracted from each facial region of multiple people. Here, the storage unit (110) can be provided in the form of a separate storage medium installed inside the identity authentication device (100), but is not limited thereto, and can also be provided in the form of an independent database (DB) connected to the same network as the identity authentication device (100). Therefore, it should be understood that the operations of the authentication unit (130), the classification unit (140), and the control unit (150) described below to access, modify, and delete feature information stored in the storage unit (110) include not only accessing, modifying, and deleting feature information in the storage medium installed inside the identity authentication device (100), but also including accessing, modifying, and deleting the database by the identity authentication device (100).

[0031] Meanwhile, the feature information stored in the storage unit (110) may be information extracted in the form of a vector value from the features including eyes, nose, mouth, and facial outlines in the facial area, and each feature information may be grouped by person unit and stored in the storage unit (110) when registering the feature information of a specific person and when classifying according to the classification of the classification unit (140) described later.

[0032] The communication unit (120) communicates with the authentication terminal (200) and can receive feature information including multiple features of the facial area generated from the photographed facial image of the person requesting identity authentication from the authentication terminal. Here, the communication unit (120) can be provided in the form of a communication module connected to a typical wired communication network and a wireless communication network. In addition, the communication unit (120) not only receives feature information about the person requesting identity authentication from the authentication terminal (200) as described above, but can also provide the identity authentication result of the authentication unit (130) to be described later to the authentication terminal (200). In addition, when the storage unit (110) is provided in the form of a database that is provided in an independent form, the authority to access, modify, and delete information stored in the database is granted, thereby supporting the authentication unit (130), the classification unit (140), and the control unit (150) to be described later.

[0033] The authentication unit (130) can check whether, among the plurality of characteristic information stored in the storage unit (110), there is characteristic information whose similarity with the characteristic information received through the communication unit (110) exceeds a preset threshold, and can transmit the identity authentication result to the authentication terminal (200) through the communication unit (110) according to the confirmation result. Here, the authentication unit (130) can perform identity authentication on the person (O) requesting identity authentication by checking whether the similarity value calculated through calculating the similarity between the characteristic information received from the communication unit (110) and each vector value corresponding to the characteristic information stored in the storage unit (110) exceeds a preset threshold.

[0034] Hereinafter, the identity authentication device (100) according to one embodiment of the present invention will be described as calculating the similarity between vector values ​​using a similarity calculation method using a cosine distance (or cosine similarity, Cosine distance), but is not limited thereto, and the identity authentication device (100) may also calculate the similarity between each vector value using any one of the Euclidean distance, the Manhattan distance, the Mahalanobis distance, and the histogram intersection, depending on the vector value generation method of the feature information.

[0035] The authentication unit (130) calculates the similarity based on the cosine distance of the following [Formula 1], and calculates the vector value V corresponding to each different feature information. a Wow V b A similarity value (S) between 0 and 1 can be calculated, where the closer the similarity value (S) is to 0, the lower the similarity, and the closer it is to 1, the higher the similarity.

[0036]

[0037] [Formula 1]

[0038]

[0039]

[0040] At this time, the authentication unit (130) calculates the similarity of each vector value between all the feature information classified and stored in the storage unit (110) by group and the feature information received from the communication unit (120), and checks whether the calculated similarity value is higher than a preset threshold value, and checks whether there is a feature point that is similar to a certain level or higher among the feature points of the face of the person received from the authentication terminal (200) and the feature points of the face of the multiple persons pre-registered and stored in the storage unit (110), thereby performing identity authentication through whether the person is pre-registered by facial recognition, and if it is confirmed that there is at least one feature information whose similarity with the feature information received from the authentication terminal (200) exceeds the threshold value among the multiple feature information stored in the storage unit (110), the authentication unit (130) notifies the authentication terminal (200) that the identity authentication of the person requesting identity authentication has been successful, or conversely, if it is confirmed that there is no feature information whose similarity with the feature information received from the authentication terminal (200) exceeds the threshold value among the multiple feature information stored in the storage unit (110). If confirmed, the authentication terminal (200) notifies that the identity authentication of the person requesting the identity authentication has failed, and the identity authentication of the person requesting the identity authentication is performed through the authentication terminal (200).

[0041] The classification unit (140) can group and classify the characteristic information received through the communication unit (120) and the characteristic information whose similarity exceeds the threshold stored in the storage unit (110) by each person unit only when the similarity is confirmed to exceed the preset threshold in the authentication unit (130). Here, when the classification unit (140) determines that the person who requested identity authentication through the authentication terminal (200) is a pre-registered person based on the result of whether the characteristic information of the authentication unit (130) exceeds the threshold, the authentication unit (130) can group, into the same group, the characteristic information received from the authentication terminal (200) through the communication unit (120) in response to the identity authentication request of the authentication terminal (200) and the characteristic information stored in the storage unit (110), the similarity with the corresponding characteristic information being confirmed to exceed the threshold.

[0042] At this time, the authentication unit (130) can classify all feature information, among the plurality of feature information stored in the storage unit (110), whose similarity with the feature information received from the communication unit (130) exceeds a threshold, into a single group, and even in the case of feature information already classified and stored as a group in the storage unit (110), the group can be not released, and only the feature information corresponding to the authentication result of the authentication unit (130) can be classified into the group. At this time, the group of characteristic information classified by the classification unit (140) is classified only by the characteristic information of the target whose identity has been successfully authenticated by the authentication unit (130), in other words, only the characteristic information confirmed as the face of a pre-registered person is classified. If the authentication unit (130) confirms that there is no characteristic information among the characteristic information stored in the storage unit (110) whose similarity with the characteristic information received from the authentication terminal (200) exceeds a threshold value, the classification unit (140) may classify the characteristic information received from the authentication terminal (200) into groups or delete it immediately without storing it in the storage unit (110).

[0043] This grouping of the characteristic information for each person in the authentication unit (130) is to confirm the number of characteristic information used for authentication for each person in order to adjust the number of characteristic information stored in the storage unit (110) to an appropriate level in order to prevent excessive time and resources (cost) from being consumed in the process of confirming the similarity between characteristic information due to an increase in the number of characteristic information stored in the storage unit (110) when the identity authentication device (100) according to the present invention accumulates and stores the target characteristic information in the storage unit (110) each time the identity authentication is successful, and this will be described in more detail in the description of the configuration of the control unit (150) to be described later.

[0044] The control unit (150) can check the number of characteristic information of each group classified and stored in the storage unit (110) by the classification unit (140) in units of groups, and can adjust the number of characteristic information of each group depending on whether the number of characteristic information of each group exceeds a preset standard number. Here, the control unit (150) can adjust the number of characteristic information of each group by comparing the similarity between the characteristic information of each group and excluding the characteristic information with the lowest similarity between the characteristic information from the corresponding group only when the number of characteristic information of each group classified by the classification unit (140) exceeds the preset standard number. In addition, the control unit (150) can check the group classification time of each characteristic information by the classification unit (140) for groups in which the number of characteristic information is equal to or greater than a preset minimum standard number at a predetermined periodic unit, and can adjust the number of characteristic information of each group by excluding the characteristic information with the oldest classification time from the corresponding group.

[0045] At this time, the similarity comparison between the feature information within the group by the control unit (150) can be performed through the following [Formula 2], which is similar to the calculation of the similarity (S) value of the vector value corresponding to each feature information by the authentication unit (130) described above.

[0046]

[0047] [Formula 2]

[0048]

[0049]

[0050] In other words, the control unit (150) maintains the number of feature information within a group to be greater than or equal to the minimum number and less than the maximum number by the classification unit (140), and by excluding the oldest feature information or the feature information with the lowest similarity between each feature information within the group based on the time of classification into the group, the number of feature information for each group that is the target of verification of whether the similarity threshold with the feature information received from the communication unit (120) of the authentication unit (130) is exceeded is maintained at a certain level, thereby preventing the time and resources required for identity authentication of the authentication unit (130) from increasing unnecessarily, and performing identity authentication based on the feature information of a relatively recent face of the identity authentication target, thereby increasing the accuracy of identity authentication.

[0051] Meanwhile, the control unit (150) may store the feature information excluded from the group in the storage unit (110) for a predetermined period of time instead of immediately deleting it, and may also create a data table in which the feature information excluded from each group and the information about the group to which the feature information previously belonged are mutually matched, and when feature information whose similarity with the feature information received through the communication unit (110) exceeds a threshold value among the feature information classified into a specific group through the authentication unit (130), the feature information stored in the storage unit (110) in a state of being excluded from the group to which the feature information belongs before the time of confirmation may be deleted from the storage unit (110).

[0052] As described above, after feature information is excluded from a specific group by the control unit (150), if the authentication unit (130) confirms that feature information exists in the group whose similarity with the feature information from the communication unit (120) exceeds a threshold value, the control unit (150) deletes the feature information excluded from the group from the storage unit (110), thereby confirming that unnecessary feature information has been excluded from the group.

[0053] In the above example, it is explained that if the authentication unit (130) confirms that identity authentication using the characteristic information within the group is successful during the deletion suspension of the characteristic information while the control unit (150) is excluding it from a specific group, the deletion suspension of the characteristic information is revoked and the characteristic information is deleted from the storage unit (110). However, if the face of the identity authentication target changes within a short period of time due to a change in the style or makeup of the identity authentication target, such as returning to a previous point in time, the authentication unit (130) may have difficulty authenticating the identity using only the latest characteristic information classified in each group or characteristic information with similar similarity.

[0054] In preparation for such a case, the authentication unit (130) does not immediately notify the authentication terminal (200) of an identity authentication failure only when the similarity between all the plurality of characteristic information classified and stored in the storage unit (110) in groups and the characteristic information received through the communication unit (120) is below a threshold value, but instead, among the plurality of characteristic information having a similarity below the threshold value with the characteristic information received through the communication unit (110), the group in which the characteristic information having the highest similarity is classified is checked, and additionally checks whether the similarity between the characteristic information excluded from the group and the characteristic information received through the communication unit (120) exceeds the threshold value through the data table generated by the control unit (150).

[0055] If, through the authentication unit (130), it is confirmed that among the feature information previously excluded from the group, there is feature information whose similarity with the feature information received through the communication unit (120) exceeds a threshold, the classification unit (140) reclassifies the feature information into the group it previously belonged to based on the data table generated from the control unit (150), so that the feature information can be used for identity authentication, so that identity authentication through the authentication unit (130) can be smoothly performed even if the face of the identity authentication target suddenly changes.

[0056]

[0057] That is, the facial recognition-based identity authentication device according to the present invention maintains a certain number of feature information, which is the target of verification of whether the similarity threshold with the feature information of the person requesting identity authentication received from the authentication terminal, in a state in which the feature information is updated for each person, thereby preventing an unnecessary increase in the time and resources required for identity authentication confirmation, and since the feature information on the relatively recent face of the identity authentication target is maintained, there is an effect of enabling rapid and highly accurate face recognition-based identity authentication.

[0058] In addition, the facial recognition-based identity authentication device according to the present invention does not immediately delete from the storage area, among the characteristic information of each person unit, the characteristic information that is oldest or has the lowest mutual similarity, but retains it until the time of identity authentication of the person in question, and then additionally utilizes it according to the result of the identity authentication of the person in question, thereby ensuring a certain level of identity authentication even when the face of the person in question changes.

[0059]

[0060] The above-described embodiments of the present invention are disclosed for the purpose of illustration, and those skilled in the art with ordinary knowledge of the present invention will be able to make various modifications, changes, and additions within the spirit and scope of the present invention, and such modifications, changes, and additions should be considered to fall within the scope of the following patent claims.

Claims

1. A storage unit in which multiple feature information extracted from each facial region of multiple people is stored; A communication unit that communicates with an authentication terminal and receives feature information including multiple features of a facial area generated from a photographed facial image of a person requesting identity authentication from the authentication terminal; An authentication unit that checks whether, among the plurality of characteristic information stored in the storage unit, there is characteristic information whose similarity with the characteristic information received through the communication unit exceeds a preset threshold, and transmits an identity authentication result to the authentication terminal through the communication unit according to the result of the check; A classification unit that groups and classifies the characteristic information received through the communication unit and the characteristic information exceeding the threshold of similarity stored in the storage unit into units for each person only when the similarity is confirmed to exceed a preset threshold in the above authentication unit; and A control unit that checks the number of characteristic information of each group classified and stored in the storage unit by the classification unit and controls the number of characteristic information in each group depending on whether the number of characteristic information in each group exceeds a preset standard number; The above authentication unit checks whether the similarity between the feature information received through the communication unit and the feature information classified into groups by the classification unit among the feature information stored in the storage unit exceeds a threshold value. A facial recognition-based identity authentication device characterized in that the control unit compares the similarity between the feature information within each group and excludes the feature information with the lowest similarity between the feature information from the group only when the number of feature information within each group classified by the classification unit exceeds a preset standard number, thereby controlling the number of feature information within each group.

2. In paragraph 1, A facial recognition-based identity authentication device characterized in that the above feature information is information extracted in the form of vector values ​​from features including eyes, nose, mouth, and facial outlines in the facial area.

3. In paragraph 1, The above control unit checks the group classification time of each characteristic information by the classification unit for a group in which the number of characteristic information is greater than or equal to a preset minimum standard number in a predetermined periodic unit, and excludes the characteristic information with the oldest classification time from the group, thereby controlling the number of characteristic information within each group. A facial recognition-based identity authentication device.

4. In paragraph 3, A facial recognition-based identity authentication device characterized in that the control unit deletes characteristic information excluded from the group in the storage unit from the storage unit after a preset period of time.

5. In paragraph 3, A facial recognition-based identity authentication device characterized in that, when the control unit confirms that the similarity between the characteristic information received through the communication unit and the characteristic information exceeds a threshold value through the authentication unit, the characteristic information excluded from the group to which the characteristic information belongs before the confirmation time is deleted from the storage unit.

6. In paragraph 5, A facial recognition-based identity authentication device characterized in that the authentication unit additionally checks whether the similarity between at least one piece of characteristic information excluded from the group by the control unit and the characteristic information received through the communication unit exceeds the threshold value only when the similarity between all the plurality of characteristic information classified and stored in the storage unit in groups and the characteristic information received through the communication unit is below the threshold value.

7. In paragraph 6, A facial recognition-based identity authentication device characterized in that the authentication unit identifies a group classified with feature information having the highest similarity among a plurality of feature information having a similarity less than a threshold value with feature information received through the communication unit, and identifies whether the similarity between at least one feature information excluded from the group by the control unit and the feature information received through the communication unit exceeds a threshold value.

8. In paragraph 7, A facial recognition-based identity authentication device characterized in that, when the classification unit confirms that, among at least one characteristic information excluded from the group through the authentication unit, the similarity between the characteristic information received through the communication unit exceeds a threshold value, the characteristic information is reclassified into an existing group.

Images