Subway gate registration and verification method and system for reducing face misrecognition rate

Abstract

This invention discloses a subway gate verification method and system to reduce the false recognition rate of facial recognition. The system includes a registration module for receiving registration information sent by a user terminal, determining whether a similar user exists in the user database, and storing the user's registration information in a first facial database if no similar user exists, and receiving registration information containing gesture information sent by the user terminal and storing it in a second facial database if a similar user exists. A verification module is also included for analyzing the received gate access information. If the analysis result shows only gate access facial information or both gate access facial information and gate access gesture information, a match is made between the first and second facial databases based on the analysis result. If a match is successful, an opening signal is sent to the gate; otherwise, a prompt is returned to the user for re-verification. This system distinguishes similar individuals during registration and stores them in different databases to avoid false recognition of users with similar faces during gate access verification, thus reducing the false recognition rate of subway gate access.

Description

Technical Field

[0001] This invention relates to the field of subway gate technology, specifically to a subway gate registration and verification method and system for reducing facial recognition misrecognition rates. Background Technology

[0002] With the application of artificial intelligence technology in commerce and security, facial recognition technology has been gradually and maturely applied to many industries. Everyday activities such as facial recognition payment and subway gate access have improved efficiency in life and work. Facial recognition technology allows users to verify their faces without contact or extra steps, which is very convenient. To facilitate subway travel, subways are also gradually adopting facial recognition technology for gate access. However, facial recognition itself has a certain false recognition rate. If two people have very similar facial features, false recognition is likely to occur, and some users have reported being recognized by other people using facial recognition. The latest existing technology can achieve a false recognition rate of one in a million when the face database is 500,000 (that is, when identifying a specific person from 500,000 people), which can meet the high accuracy requirements of facial recognition. However, as the number of users using facial recognition to pass through gates increases, the face database is constantly increasing. If the database reaches 20 million, the false recognition rate will theoretically increase to one in ten thousand. For example, among 20 million people, there is a high probability that there will be two or even more people who look very similar. In this case, false recognition is likely to occur. Therefore, how to reduce the facial recognition rate in subway gates is a problem that needs to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a subway gate registration and verification method and system that reduces the false recognition rate of faces. For similar users, face recognition and gesture recognition are combined to perform hybrid recognition, which greatly reduces the false recognition rate of face recognition.

[0004] To achieve the above objectives, this application proposes the following solution:

[0005] On the one hand, the present invention provides a subway gate registration method to reduce the rate of facial recognition errors, specifically including the following steps:

[0006] S100: Receive the first registration information sent by the user terminal of the new user, and analyze the first registration information to obtain the registered face information;

[0007] S110. Compare and match the registered facial information against the user database formed by registered users.

[0008] If no matching face is found, the registered face information will be stored in the first face database.

[0009] If a face is matched, the user corresponding to that face is marked as a similar user, and a prompt to add gesture recognition is sent to the currently registered user's terminal;

[0010] S120: Receive the second registration information sent by the user terminal of the new user. The second registration information includes facial information and gesture information. Store the second registration information in the second facial database.

[0011] In some specific implementations, the steps also include:

[0012] S130. After the current user completes registration, if a similar user is in the first face database, the similar user will be deleted from the first face database, a prompt to add gesture recognition will be sent to the user's client, the updated registration information sent by the similar user will be received, and the updated registration information will be stored in the second face database.

[0013] In some specific implementations, when the updated registration information sent by the user is received, the updated registration information is identified, the updated facial information and the updated gesture information are identified, and the updated gesture information is compared with the gesture information of the new user corresponding to the similar user in the second face database. If they match, a prompt to change the recognition gesture is returned, and the updated facial information and the updated gesture information are bound together to form a recognition data pair and stored in the second face database.

[0014] In some specific implementations, in step S100, the first registration information includes multiple face images and identity authentication information. Face features are extracted from the multiple face images respectively, and the face image with the highest face feature value is used as the registered face information. The multiple face images are then placed into the same array in descending order of feature value and stored in the user database.

[0015] Secondly, this application provides a subway gate verification method to reduce the false recognition rate of faces, including the following steps:

[0016] S200: Receive the current user's gate passage information uploaded by the gate passage device;

[0017] S210. Analyze the gate passage information to obtain the analysis results. The analysis results are either only containing gate passage face information or simultaneously containing gate passage face information and gate passage gesture information.

[0018] S220. Based on the analysis results, call the first face database and the second face database, and compare the analysis results in the first face database and the second face database. If the match is successful, send an opening signal to the gate; if the match is not successful, return a prompt to remind the user to re-verify.

[0019] In some specific implementations, when the analysis result in step S220 only contains the face information of the person passing through the gate, the first face database is called to match the face information of the person passing through the gate in the first face database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned to remind the user to re-verify.

[0020] In some specific implementations, when the analysis results contain both the face information and gesture information of the person passing through the gate, a second face database is called to determine whether the face information and gesture information can be successfully matched in the second face database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify.

[0021] In some specific implementations, when the gate access information does not match in the second face database, the following steps are also included:

[0022] If the facial information of the person passing through the gate matches, but the gesture information does not match in the second facial database, a prompt will be returned, asking the user to re-verify.

[0023] If the facial information of the person passing through the gate does not match, the first facial database is called to match the facial information of the person passing through the gate with the first facial database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify.

[0024] Thirdly, this application provides a subway turnstile system to reduce the false recognition rate of facial recognition, including a user terminal, a server, and turnstile equipment. The server includes a registration module and a verification module, and the turnstile equipment includes a turnstile and a PAD installed on the turnstile.

[0025] The registration module is used to receive registration information sent by the user terminal, determine whether there is a similar user in the user database based on the registration information, and if not, store the user's registration information in the first face database. If it does, it receives registration information containing gesture information sent by the user terminal and stores the user's registration information in the second face database.

[0026] The verification module receives the gate passage information uploaded by the PAD scan, analyzes the information, and obtains the analysis results. The analysis results may contain only the gate passage face information or both the gate passage face information and the gate passage gesture information. Based on the analysis results, the module matches the information against the first face database and the second face database. If the match is successful, the module sends an opening signal to the gate; if the match is not successful, the module returns a prompt, asking the user to re-verify.

[0027] In some specific implementations, an update module is also included, which is used to determine whether a similar user is in the first face database after the current user has registered. If so, the similar user is deleted from the first face database, and a prompt to add gesture recognition is sent to the user terminal of the similar user. The module also receives the updated registration information sent by the similar user and stores the updated registration information in the second face database.

[0028] The beneficial effects of this invention are as follows:

[0029] This application sets up three databases: a user database, which includes all facial information of all registered users; a first facial database, which includes registered users who only registered their facial information during registration; and a second facial database, which includes registered users who registered both their facial and gesture information during registration. Therefore, when a new user registers, the new user's first registration information is first compared and matched against the user database to determine if a match can be found. If a match is found, it means that there is a user with a similar face. At this time, the user is prompted to set up gesture + facial information, and the second registration information of gesture + facial information is sent to the server. The server saves the new user's second registration information to the second facial database. If no match is found, it means that there is no user with a similar face, and the new user's first registration information is stored in the first facial database.

[0030] When a user enters the station through the turnstile, the subway turnstile equipment collects the information of the person entering the station and identifies the information of the person passing through the turnstile. If only facial information is included, it is matched against the first facial database. If the match is successful, the turnstile allows passage. If both facial and gesture information are included, it is matched against the second facial database. If the match is successful, the turnstile allows passage. If the gesture is incorrect, a re-verification prompt is made. If the facial information is incorrect, it is compared against the first facial database. If a match is found, the turnstile allows passage. If a match is not found, re-verification is made.

[0031] This application distinguishes similar individuals during registration and stores them in different databases, thus avoiding misidentification of users with similar faces during gate verification, reducing the misidentification rate of subway gates, and reminding users of the possibility of misidentification during user registration, thereby improving users' security awareness. Attached Figure Description

[0032] Figure 1 This is a block diagram of a subway gate system for reducing face misrecognition rate provided in an embodiment of the present invention;

[0033] Figure 2 This is a flowchart of a subway gate registration method for reducing face misrecognition rate provided in an embodiment of the present invention;

[0034] Figure 3 This is a flowchart of a subway gate verification method for reducing face false recognition rate provided in an embodiment of the present invention;

[0035] Figure 4 The diagram illustrates three types of databases provided in the embodiments of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention.

[0038] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0039] Furthermore, for clarity and brevity, descriptions of well-known structures, functions, and configurations may have been omitted. Those skilled in the art will recognize that various changes and modifications can be made to the examples described herein without departing from the spirit and scope of this disclosure.

[0040] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0041] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0042] Example

[0043] like Figure 1 As shown in the figure, this embodiment provides a subway gate system for reducing the false recognition rate of faces, including a user terminal, a server, and gate equipment. The server includes a registration module and a verification module. The gate equipment includes a gate and a PAD installed on the gate. The face recognition camera of the gate PAD recognizes the passenger's face, and the panoramic camera recognizes the passenger's gestures. The passenger is recognized and enters the station.

[0044] The registration module is used to receive registration information sent by the user terminal, determine whether there is a similar user in the user database based on the registration information, and if not, store the user's registration information in the first face database. If it does, it receives registration information containing gesture information sent by the user terminal and stores the user's registration information in the second face database.

[0045] The verification module receives the gate passage information uploaded by the PAD scan, analyzes the information, and obtains the analysis results. The analysis results may contain only the gate passage face information or both the gate passage face information and the gate passage gesture information. Based on the analysis results, the module matches the information against the first face database and the second face database. If the match is successful, the module sends an opening signal to the gate; if the match is not successful, the module returns a prompt, asking the user to re-verify.

[0046] The update module is used to determine whether a similar user is in the first face database after the current user has registered. If so, the similar user is deleted from the first face database, and a prompt to add gesture recognition is sent to the user's client. The module also receives the updated registration information sent by the similar user and stores the updated registration information in the second face database.

[0047] Specifically, for each module, this embodiment provides a specific implementation method, wherein, as... Figure 2 As shown, for the registration module, a method for subway gate registration to reduce the rate of false facial recognition specifically includes the following steps:

[0048] S100: Receive the first registration information sent by the new user's client; analyze the first registration information to obtain the registered facial information; the new user registers for facial recognition and activates the facial recognition function on the client.

[0049] S110. The registered facial information is compared and matched in the user database formed by registered users. The first registration information includes multiple facial images and identity authentication information. Facial features are extracted from the multiple facial images respectively. The facial image with the highest facial feature value is used as the registered facial information. The multiple facial images are placed into the same array in descending order of feature value and stored in the user database. The user database is bound to the first facial database and the second facial database. Whenever a new user registers, data will be added to the user database.

[0050] Specifically, the comparison mentioned here can be understood as comparing the passenger's facial information with the user database. When the similarity of the facial information reaches a certain threshold, it is considered that the face matches the face in the user database. The extraction of facial information (facial features) and the similarity recognition of two faces are existing technologies and will not be described in detail in this application. If the faces of multiple different users are matched, all of these users are marked as similar users. When a new user registers, a reminder is sent to similar users so that people who may be misidentified are all given different gesture recognitions to avoid misidentification between similar users.

[0051] If a new user's face information does not match in the user database, it means that the probability of the user being misidentified is low, and the registered face information will be stored in the first face database.

[0052] If a new user's face is matched in the user database, the user corresponding to that face will be marked as a similar user, and a prompt to add gesture recognition will be sent to the currently registered user's terminal.

[0053] S120: Receive the second registration information sent by the user terminal of the new user. The second registration information includes facial information and gesture information. Store the second registration information in the second facial database.

[0054] In step S120, it is necessary to determine whether the similar user comes from the first face database or the second face database. If it comes from the first face database, the similar user needs to perform the update operation in step S130. If it comes from the second face database, the gesture information in the second registration information needs to be compared with the gesture information of the similar user. If they are the same, the new user needs to be reminded to set different gestures. If they are different, the second registration information is stored in the second face database.

[0055] When a similar user is in the first face database, the new user is prompted to select a gesture from the set of commonly used gestures or to set a private gesture. When a similar user is in the second face database, the new user is prompted to set a private gesture first. The gesture selected or set by the new user is compared with the gesture of the similar user to see if they are the same or similar (e.g., compared to one or two). If they are the same or similar, the new user is prompted to change the gesture until it is approved.

[0056] To improve recognition speed, a gesture library is set up on the user side, which includes a set of commonly used gestures and a set of private gestures. When registering on the user side, users can choose a gesture from the set of commonly used gestures to register. Users can also set unique gestures and save them to the set of private gestures. In this way, if there are similar users, it can detect whether the difference between the user's gestures is difficult to identify. If so, it can remind the new user to set private gestures or push gestures with greater differences to the new user based on the gestures of similar users.

[0057] If similar users appear after a new user registration is completed, the system needs to be updated. The update method is as follows:

[0058] S130. After the current user completes registration, if a similar user is in the first face database, the similar user will be deleted from the first face database, a prompt to add gesture recognition will be sent to the user's client, the updated registration information sent by the similar user will be received, and the updated registration information will be stored in the second face database.

[0059] When the server receives updated registration information from a similar user, it identifies the updated facial information and updated gesture information. To avoid the similar user's gestures being similar to the new user's gestures, the updated gesture information is compared with the gesture information of the new user corresponding to the similar user in the second face database. If the two gestures are the same or similar, they are considered a match, and a prompt to change the recognition gesture is returned to the similar user's client. The finally obtained updated facial information and updated gesture information are bound together to form a recognition data pair and stored in the second face database.

[0060] For the verification module, the verification method given in this embodiment is as follows: Figure 3 As shown, it includes the following steps:

[0061] S200: Receive the current user's gate passage information uploaded by the gate passage device;

[0062] S210. Analyze the gate passage information to obtain the analysis results. The analysis results are either only containing gate passage face information or simultaneously containing gate passage face information and gate passage gesture information.

[0063] S220. Based on the analysis results, call the first face database and the second face database, and compare the analysis results in the first face database and the second face database. If the match is successful, send an opening signal to the gate; if the match is not successful, return a prompt to remind the user to re-verify.

[0064] Specifically, since users with similar characteristics are aware of each other's existence, they generally use hand gestures when passing through the gate, while ordinary users simply use facial recognition. Therefore, when the analysis results only contain facial information of those passing through the gate, the first facial database is called to match the facial information of those passing through the gate. If a match is found, an opening signal is sent to the gate; if no match is found, a prompt is returned, asking the user to re-verify.

[0065] When the analysis results contain both facial information and gesture information of the person passing through the gate, the second facial database is called to determine whether the facial information and gesture information can be successfully matched in the second facial database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify.

[0066] To avoid situations where a user belongs to the second face database but forgets to use gesture information, resulting in a matching face information for gate access but a mismatch in the second face database for gate access gesture information, a prompt is returned to the PAD, prompting the user to re-verify.

[0067] To avoid situations where a user belongs to the first face database but accidentally makes a gesture that is recognized, resulting in a mismatch between the face information and the gesture (which may or may not match), the first face database is called to match the user's face information against it. If a match is found, an opening signal is sent to the gate; otherwise, a prompt is returned asking the user to re-verify.

[0068] Generally, the false recognition rates of face recognition and gesture recognition in user identification methods are shown in the table below:

[0069]

[0070] This application combines facial recognition and gesture recognition to add identification features for users with similar faces, reducing the false recognition rate. When a new user registers, their face is captured using a liveness detection method. When the new user uploads their facial and identity information to the registration server, multiple faces are captured during liveness detection, and all of these faces are stored in the user database. The face with the highest facial feature value can then be selected as the registration face information. Figure 4 As shown, this application sets up three databases: a user database, which includes the facial information of all registered users; a first facial database, which includes registered users who only registered their facial information during registration; and a second facial database, which includes registered users who registered both their facial and gesture information during registration. Therefore, when a new user registers, the new user's first registration information is first compared and matched against the user database to determine if a match can be found. If a match is found, it means that a similar user exists, and the user is prompted to set up gesture + facial information. The second registration information, consisting of gesture and facial information, is then sent to the server, which saves the new user's second registration information to the second facial database. If no match is found, it means that no similar user exists, and the new user's first registration information is stored in the first facial database.

[0071] At the subway gate verification terminal, the system receives the gate information uploaded by the gate device, identifies the information contained in the gate information, and if it only includes facial gate information, it calls the first facial database and matches the facial gate information in the first facial database. If the match is successful, it sends an opening signal to the gate; if the match is not successful, it returns a prompt, prompting the user to re-verify.

[0072] If the system contains both facial recognition and gesture recognition information, it calls the second facial recognition database to determine if both information can be successfully matched. If a match is found, an opening signal is sent to the gate. If the facial recognition matches but the gesture recognition does not, a prompt is returned asking the user to re-verify. If the facial recognition does not match, the system calls the first facial recognition database to match the facial recognition information. If a match is found, an opening signal is sent to the gate. If a match is not found, a prompt is returned asking the user to re-verify.

[0073] It should be noted that the focus of this application is on building three databases and registration verification logic to help the server reduce the false recognition rate of similar users. The facial feature extraction, matching technology, facial recognition, gesture recognition and other technologies mentioned above can all be understood as existing technologies that are easy for those skilled in the art to implement, and will not be described in detail in this application.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Based on the technical essence of the present invention, any simple modifications, equivalent substitutions, and improvements made to the above embodiments within the spirit and principles of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A subway turnstile system for reducing facial recognition false recognition rate, characterized in that, It includes a user terminal, a server, and gate access equipment. The server includes a registration module and a verification module. The gate access equipment includes the gate itself and a PAD installed on the gate. The registration module is used to receive registration information sent by the user terminal, determine whether there is a similar user in the user database based on the registration information, and if not, store the user's registration information in the first face database. If it does, it receives registration information containing gesture information sent by the user terminal and stores the user's registration information in the second face database. The verification module receives the gate passage information uploaded by the PAD scan, analyzes the information, and obtains the analysis results. The analysis results may contain only the gate passage face information or both the gate passage face information and the gate passage gesture information. Based on the analysis results, the module matches the information against the first face database and the second face database. If the match is successful, the module sends an opening signal to the gate; if the match is not successful, the module returns a prompt to the user to re-verify. When the analysis results only contain facial information of people passing through the gate, the first facial database is called to match the facial information of people passing through the gate in the first facial database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify. When the analysis results contain both the face information and gesture information of the person passing through the gate, the second face database is called to determine whether the face information and gesture information can be successfully matched in the second face database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify. The update module is used to determine whether a similar user is in the first face database after the current user has registered. If so, the similar user is deleted from the first face database, and a prompt to add gesture recognition is sent to the user's client. The module also receives the updated registration information sent by the similar user and stores the updated registration information in the second face database.

2. A method for subway gate registration that reduces facial recognition false recognition rate, characterized in that, The application of this method in a subway turnstile system for reducing facial recognition false recognition rates, as described in claim 1, specifically includes the following steps: S100: Receive the first registration information sent by the user terminal of the new user, and analyze the first registration information to obtain the registered face information; S110. Compare and match the registered facial information against the user database formed by registered users. If no matching face is found, the registered face information will be stored in the first face database. If a face is matched, the user corresponding to that face is marked as a similar user, and a prompt to add gesture recognition is sent to the currently registered user's terminal; S120: Receive the second registration information sent by the user terminal of the new user. The second registration information includes facial information and gesture information. Store the second registration information in the second facial database.

3. The subway gate registration method for reducing facial recognition false recognition rate according to claim 2, characterized in that, It also includes the following steps: S130. After the current user completes registration, if a similar user is in the first face database, the similar user will be deleted from the first face database, a prompt to add gesture recognition will be sent to the user's client, the updated registration information sent by the similar user will be received, and the updated registration information will be stored in the second face database.

4. The subway gate registration method for reducing facial recognition false recognition rate according to claim 3, characterized in that, When an updated registration information is received from a user, the updated registration information is identified, including the updated facial information and the updated gesture information. The updated gesture information is then compared with the gesture information of the new user corresponding to the similar user in the second facial database. If a match is found, a prompt to change the recognition gesture is returned. The updated facial information and the updated gesture information are then bound together to form a recognition data pair, which is stored in the second facial database.

5. A subway gate registration method for reducing facial recognition false recognition rate according to claim 2, characterized in that, In step S100, the first registration information includes multiple face images and identity authentication information. Face features are extracted from the multiple face images respectively. The face image with the highest face feature value is used as the registered face information. The multiple face images are placed into the same array in descending order of feature value and stored in the user database.

6. A method for verifying subway gate access to reduce the false recognition rate of facial recognition, characterized in that, The application of a subway gate system for reducing facial recognition false recognition rate as described in claim 1 includes the following steps: S200: Receive the current user's gate passage information uploaded by the gate passage device; S210. Analyze the gate passage information to obtain the analysis results. The analysis results are either only containing gate passage face information or simultaneously containing gate passage face information and gate passage gesture information. S220. Based on the analysis results, call the first face database and the second face database, and compare the analysis results in the first face database and the second face database. If the match is successful, send an opening signal to the gate; if the match is not successful, return a prompt to remind the user to re-verify.

7. A subway gate verification method for reducing face false recognition rate according to claim 6, characterized in that, In step S220, if the analysis result only contains the face information of the person passing through the gate, the first face database is called to match the face information of the person passing through the gate in the first face database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned to prompt the user to re-verify.

8. A subway gate verification method for reducing face false recognition rate according to claim 6, characterized in that, When the analysis results contain both facial information and gesture information of the person passing through the gate, the second facial database is called to determine whether the facial information and gesture information can be successfully matched in the second facial database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify.

9. A subway gate verification method for reducing face false recognition rate according to claim 8, characterized in that, When the gate access information does not match in the second face database, the following steps are also included: If the facial information of the person passing through the gate matches, but the gesture information does not match in the second facial database, a prompt will be returned, asking the user to re-verify. If the facial information of the person passing through the gate does not match, the first facial database is called to match the facial information of the person passing through the gate with the first facial database. If the match is successful, an opening signal is sent to the gate; if the match is not successful, a prompt is returned, prompting the user to re-verify.

Images