Face recognition method and system compatible with multiple versions of face features

Abstract

This invention relates to the field of facial recognition technology, and particularly to a facial recognition method and system compatible with multiple versions of facial features. The method includes: extracting facial features when a passenger passes through an entrance gate and marking the SDK version number of the entrance gate in the facial features; performing identity verification using a facial registration server with the same SDK version, saving the verification results, and performing a server update process; extracting facial features when a passenger passes through a security checkpoint and marking the SDK version number of the security checkpoint in the facial features; and performing facial recognition using a facial recognition server with the same SDK version. This invention enables users to register on different versions of clients and perform recognition on different versions of servers, achieving a smooth upgrade and iteration effect for the system.

Description

Technical Field

[0001] This invention relates to the field of facial recognition technology, and in particular to a facial recognition method and system compatible with multiple versions of facial features. Background Technology

[0002] Currently, many airports have facial recognition systems. Before passengers enter the security checkpoint, there is a waiting area where passengers queue to enter the checkpoint. Before entering the waiting area, to separate non-flight passengers, their identities are verified. Additionally, to enable passengers to use the facial recognition system for security checks and boarding, their facial information is collected and registered in the facial database. This system is used, but not exclusively, through on-site entrance gates or mobile apps.

[0003] In the process of using passenger facial features, with the continuous development of technology and the constant updates to facial feature extraction methods, incompatibility between new and old versions of facial features is inevitable. To maintain system stability, partial updates to devices and servers are implemented, such as canary releases, rolling releases, and gray-scale releases. When a passenger is about to enter the restricted area and needs to pass through security, it is uncertain which version of facial features the passenger registered with in the database (entrance gate or mobile registration). At this time, it is not possible to use the facial feature version number captured by the security gate to send it to the corresponding facial recognition server through traffic redirection or redirection, because the facial recognition server may not have the corresponding facial feature version for that user. The reason is that the version number of the passenger's registered feature database does not correspond to the version number of the facial feature captured by the security gate. Summary of the Invention

[0004] To address the above problems, this invention provides a face recognition method and system compatible with multiple versions of facial features.

[0005] In a first aspect, the present invention proposes a face recognition method compatible with multiple versions of facial features, executed by a security inspection system located in an airport security checkpoint. The security inspection system includes an entrance gate, a security gate, a face registration server, and a face recognition server. The face recognition method includes the following steps:

[0006] S1. When a passenger passes through any entrance gate IG n When, where n = 1, 2, ..., N, N is the number of entrance gates, including:

[0007] S11. Entrance Gate IG n After acquiring the passenger's facial image, a registration verification request is sent; the registration verification request includes facial features extracted from the facial image and the entrance gate's IG.n The SDK version number used for extracting facial features is Ver. n And passenger user information;

[0008] S12. The gateway receives the registration verification request information and, based on the SDK version number Ver... n Forward it to the face registration server using the same SDK version number for 1:1 authentication;

[0009] S13. After successful identity verification, the passenger's facial data is added to the feature table of the database, and a corresponding record ID is added; the server update process is executed; the facial data includes user information, facial features, and SDK version number Ver. n ;

[0010] S2. When a passenger passes through any security gate SG m When, m = 1, 2, ..., M, M is the number of security gates, including:

[0011] S21. Security Gate SG m After acquiring the passenger's facial image, a facial recognition request is sent; the facial recognition information includes facial features extracted from the facial image and the security gate SG. m The SDK version number used for extracting facial features is Ver. m ;

[0012] S22. The gateway receives the face recognition information and, according to the SDK version number Ver... m Relocation is performed to forward the face recognition information to a face recognition server using the same SDK version number for face recognition.

[0013] Furthermore, the security system layout in the airport security check area is as follows: multiple independent security check lanes are set up in the airport security check area, and a security gate is set up in front of each security check lane. All security gates are arranged in parallel. At the entrance of the airport security check area, multiple entrance gates are set up in parallel, and a passenger waiting area is formed between the entrance gate queue and the security gate queue.

[0014] Furthermore, after saving the passenger's facial data, step S13 executes a server update process, including:

[0015] S131. Face Registration Server (FRS) i Determine if a Face Registration Server (FRS) exists that has a different SDK version number than the one you are using. i’ If it exists, proceed to step S132; where i, i' = 1, 2, ..., N ig And i≠i'; N ig Number of servers used for face registration;

[0016] S132. Face Registration Server (FRS) j Face Registration Server (FRS) i’ Send a message to retrieve the data;

[0017] S133. Face Registration Server (FRS) i’ The system receives the request to extract features from the passenger's face image and obtains the data from the Face Registration Server (FRS). i’ The face data with the SDK version number is added to the feature table of the database, and the corresponding record ID is added.

[0018] S134. An update message is sent to all face recognition servers using a message broker. The update message includes two combined pieces of information, one of which includes the Face Registration Server (FRS). i SDK version information, Face Registration Server (FRS) i The record ID in the feature table for the facial data extracted from the facial image; another piece of combined information includes the Face Registration Server (FRS). i’ SDK version information, Face Registration Server (FRS) i’ The record ID in the feature table for the facial data extracted from this facial image;

[0019] S135. Each face recognition server reads the corresponding combination information from the update message based on its own SDK version number.

[0020] In a second aspect, based on the method proposed in the first aspect, the present invention provides a face recognition system compatible with multiple versions of face features, comprising:

[0021] The entrance gate acquisition module is used to acquire facial images from the entrance gate, extract facial features from the facial images, and then send registration verification request information to the facial registration server.

[0022] The gateway relocation module is used to send the registration verification request information to the corresponding face registration server and the face recognition request information to the corresponding face recognition server according to the SDK version number.

[0023] The identity verification module is used by the face registration server to verify the identity of passengers based on the received registration verification request information and to save the face data of passengers who have passed the identity verification.

[0024] The server update module is used to detect the SDK version number of the face registration server and obtain face data with different SDK version numbers based on the face image of the same passenger.

[0025] The security gate data acquisition module is used to acquire facial images from the security gate, extract facial features from the facial images, and then send facial recognition request information to the facial recognition server.

[0026] The face recognition module is used by the face recognition server to perform face recognition based on the received face recognition request information.

[0027] Furthermore, the facial recognition system also includes:

[0028] The information registration module is used by passengers to collect facial images using the terminal device APP, and then send the facial image, the facial features extracted from the facial image, the SDK version number used by the terminal device APP to extract the facial features, and the passenger's user information to the facial registration server.

[0029] The beneficial effects of this invention are:

[0030] The method proposed in this invention marks the extracted facial features with SDK version information during system upgrades. During recognition and comparison, it utilizes request redirection technology based on the SDK version information to accurately identify the corresponding feature version. This allows users to register on different client versions and be recognized on different server versions, achieving a smooth system upgrade and iteration effect.

[0031] Facial feature extraction is a very time-consuming and CPU-intensive task. The more channels there are, the greater the pressure on the backend server. To meet performance requirements, the backend server needs continuous hardware upgrades, and when the hardware is insufficient, server expansion is necessary. Therefore, this invention, based on the concept of edge computing, allows the client to share the burden of the backend server by directly using a gate for facial feature extraction, thus eliminating the need for the backend server to perform facial feature extraction. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of an airport security checkpoint area according to an embodiment of the present invention;

[0033] Figure 2 This is a flowchart of the method framework of the present invention;

[0034] Figure 3 This is a network topology diagram of the system of the present invention;

[0035] Figure 4 This is a flowchart of the method 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. 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] Currently, almost all airports have introduced facial recognition systems to assist in security checks. However, with technological advancements, these systems require continuous updates. During the update process of the facial recognition SDK, new and old versions often coexist, and the facial features collected by the two versions may be incompatible. This poses new challenges to the upgrade and stability of the facial recognition system. To eliminate the impact of version incompatibility during the facial recognition SDK update process and ensure that passengers can pass through any gate by facial recognition, this invention provides a facial recognition method compatible with multiple versions of facial features. This method is executed by a security system located in the airport security area, which includes entrance gates, security gates, a facial registration server, and a facial recognition server. Specifically, in this invention, passengers can register their faces by choosing a terminal device app or by having their images captured by the gate. The front end extracts facial features using the facial SDK and sends them to the backend server for database construction. The backend server includes a gateway, a facial registration server, and a facial recognition server, all of which exist in a cluster.

[0038] Specifically, the security system layout in the airport security check area is as follows: multiple independent security check lanes are set up in the airport security check area, and a security gate is set up in front of each security check lane. All security gates are arranged in parallel. At the entrance of the airport security check area, multiple entrance gates are set up in parallel, and a passenger waiting area is formed between the entrance gate queue and the security gate queue.

[0039] Based on the above conditions, such as Figure 4 As shown, the face recognition method includes the following steps:

[0040] S1. When a passenger passes through any entrance gate IG n When, where n = 1, 2, ..., N, N is the number of entrance gates, including:

[0041] S11. Entrance Gate IG n After acquiring the passenger's facial image, a registration verification request is sent; the registration verification request includes facial features extracted from the facial image and the entrance gate's IG. n The SDK version number used for extracting facial features is Ver. nAnd passenger user information; S12. The gateway receives the registration verification request information and, according to the SDK version number Ver n Forward it to the face registration server using the same SDK version number for 1:1 authentication;

[0042] S13. After successful identity verification, the passenger's facial data is added to the feature table of the database, and a corresponding record ID is added; the server update process is executed; the facial data includes user information, facial features, and SDK version number Ver. n ;

[0043] Specifically, there are two ways to determine whether to execute the server update process. One is to determine whether the current state is one of simultaneous releases with different SDK version numbers through global configuration; the other is to determine the SDK version information of each server through the metadata of each server in the registry center.

[0044] S2. When a passenger passes through any security gate SG m When, m = 1, 2, ..., M, M is the number of security gates, including:

[0045] S21. Security Gate SG m After acquiring the passenger's facial image, a facial recognition request is sent; the facial recognition information includes facial features extracted from the facial image and the security gate SG. m The SDK version number used for extracting facial features is Ver. m ;

[0046] S22. The gateway receives the face recognition information and, according to the SDK version number Ver... m Relocation is performed to forward the face recognition information to a face recognition server using the same SDK version number for face recognition.

[0047] The reason this invention utilizes turnstiles for feature extraction instead of a backend server is that facial feature extraction is a very CPU-intensive task. The more channels there are, the greater the pressure on the backend server. To meet performance requirements, the server needs continuous hardware upgrades, and when the hardware is insufficient, continuous server expansion is necessary. This approach adopts the concept of edge computing, and since the client has abundant spare computing power, using the client to extract facial features alleviates the server's workload.

[0048] Specifically, after saving the passenger's facial data, step S13 executes a server update process, including:

[0049] S131. Face Registration Server (FRS) iDetermine if a Face Registration Server (FRS) exists that has a different SDK version number than the one you are using. i’ If it exists, proceed to step S132; where i, i' = 1, 2, ..., N ig And i≠i'; N ig Number of servers used for face registration;

[0050] S132. Face Registration Server (FRS) i Face Registration Server (FRS) i’ Send a message to retrieve the data;

[0051] S133. Face Registration Server (FRS) i’ The system receives a request to extract facial features from the passenger's face image to obtain new facial data. This facial data includes user information, facial features, and the Face Registration Server (FRS). i’ The SDK version number; add the new face data to the feature table of the database and add the corresponding record ID;

[0052] S134. An update message is sent to all face recognition servers using a message broker. The update message includes two combined pieces of information, one of which includes the Face Registration Server (FRS). i SDK version information, Face Registration Server (FRS) i The record ID in the feature table for the facial data extracted from the facial image; another piece of combined information includes the Face Registration Server (FRS). i’ SDK version information, Face Registration Server (FRS) i’ The record ID in the feature table for the facial data extracted from this facial image;

[0053] S135. Each face recognition server reads the corresponding combination information in the update message according to its own SDK version number, loads the corresponding data from the database according to the record ID in the combination information, and then updates the face feature database information in its own memory.

[0054] Specifically, in step S131, if there is a match with the face registration server FRS... i If multiple face registration servers use different SDK versions, and these multiple face registration servers also use different SDK versions, then the face registration server FRS... i The system sends retrieval requests to each of the multiple face registration servers, and then executes steps S133-S135. When the multiple face registration servers use the same SDK version, the face registration server FRS... i Send a call to retrieve a message to one of the face registration servers, and then execute steps S133-S135.

[0055] In one embodiment, the security screening system installed in the airport security check area is as follows: Figure 1 As shown, the system includes 3 entrance gates and 4 security gates. The SDK version for each entrance gate and security gate is updated independently, resulting in the coexistence of both old version V1 and new version V2. Furthermore, in the passenger waiting area formed between the entrance and security gates, there are no one-to-one independent channels between the security and entrance gates. Therefore, which security gate a passenger will approach after entering the waiting area is random. This can lead to incompatibility between the SDK version used for facial feature collection at the entrance gates and the SDK version used at the security gates. To eliminate the impact of version incompatibility, the facial recognition method proposed in this invention is used, specifically including:

[0056] like Figure 1 , Figure 2 As shown, when a passenger passes through the intermediate entrance gate, the gate captures the passenger's face using a camera and selects the best image. The intermediate entrance gate uses SDK version V2, so it uses this version to extract facial features from the image. Then, the gate sends a registration verification request to the face registration server, consisting of the facial features, its SDK version number (V2), and the passenger's user information. Upon receiving the request, the face registration server begins a 1:1 identity verification process. If successful, the passenger can proceed to the waiting area; otherwise, the gate needs to re-capture the facial features and resend the request. To avoid issues caused by SDK version incompatibility, the gateway redirects the registration verification request sent by the intermediate entrance gate. Specifically, the gateway forwards the request to the face registration server using the same SDK version number for 1:1 identity verification. After identity verification, the passenger's facial data is saved in the database and a corresponding record ID is added; at the same time, the server update process is executed; the facial data includes user information, facial features, and SDK version number V2.

[0057] When a passenger passes through any security gate, the gate captures an image of the passenger's face using its camera and extracts facial features. If the security gate uses SDK version V2, the facial features correspond to SDK version V2, and the sent facial recognition request includes the facial features and the SDK version V2 used by the security gate. If the security gate uses SDK version V1, the facial features correspond to SDK version V1, and the sent facial recognition information includes the facial features and the SDK version V1 used by the security gate. The gateway receives the facial recognition information and, based on the SDK version number, relocates the information and forwards it to a facial recognition server using the same SDK version number for facial recognition.

[0058] Through the above process, passengers can have their facial images captured and registered in the database on any client device. When upgrading the client and server, there's no need to consider facial feature version issues. A detailed flowchart can be found here. Figure 2 As shown.

[0059] This approach allows for the use of common techniques such as rolling deployment, blue-green deployment, canary deployment, and parallel deployment when iterating and upgrading the feature versions of the face recognition SDK on the client or server. This effectively avoids the problem of users registering on different versions and being unable to be recognized in other versions during the upgrade process.

[0060] For long-term facial recognition databases, such as those used in residential community access control systems, the collected facial data is stored in a database indefinitely. Upgrading features requires upgrading both the client and server, and migrating the data for the new features. This method, however, only requires extracting the new version of features from historical facial data. The backend facial recognition server maintains both versions running, and new users are registered using both versions for feature extraction. During recognition, the feature identifier is used to redirect data to the corresponding facial server for identification. Once the system is stable and the client upgrade is complete, all backend servers are replaced, and the old facial features are deleted.

[0061] For a detailed network topology diagram, please refer to... Figure 3 As shown, the entire system consists of an entrance gate, a security gate, a gateway, a registration server, and an identification server. During upgrades and updates, the gateway plays a crucial role in redirecting front-end requests, ensuring consistency in feature versions between the front-end and back-end, and performing feature comparison and identification.

[0062] In one embodiment, the present invention also provides a face recognition system compatible with multiple versions of face features, comprising:

[0063] The entrance gate acquisition module is used to acquire facial images from the entrance gate, extract facial features from the facial images, and then send registration verification request information to the facial registration server.

[0064] The gateway relocation module is used to send the registration verification request information to the corresponding face registration server and the face recognition request information to the corresponding face recognition server according to the SDK version number.

[0065] The identity verification module is used by the face registration server to verify the identity of passengers based on the received registration verification request information and to save the face data of passengers who have passed the identity verification.

[0066] The server update module is used to detect the SDK version number of the face registration server and obtain face data with different SDK version numbers based on the face image of the same passenger.

[0067] The security gate data acquisition module is used to acquire facial images from the security gate, extract facial features from the facial images, and then send facial recognition request information to the facial recognition server.

[0068] The face recognition module is used by the face recognition server to perform face recognition based on the received face recognition request information.

[0069] Specifically, the facial recognition system also includes:

[0070] The information registration module is used by passengers to collect facial images using the terminal device APP, and then send the facial image, the facial features extracted from the facial image, the SDK version number used by the terminal device APP to extract the facial features, and the passenger's user information to the facial registration server.

[0071] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "setting," "connection," "fixing," "rotation," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal connection of two components or the interaction between two components. Unless otherwise explicitly limited, those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0072] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A face recognition method compatible with multiple versions of face features, characterized in that, The face recognition method is executed by a security check system arranged in an airport security check area, and the security check system comprises an entrance gate, a security check gate, a face registration server and a face recognition server. S1. When the passenger passes through any entrance gate IG n wherein n = 1, 2, …, N, N is the number of entrance gates, comprising: S11. Entrance gate IG n After obtaining the face image of the passenger, a registration and verification request information is sent; the registration and verification request information includes face features extracted according to the face image, an SDK version number Ver used for extracting the face features, and user information of the passenger. n After obtaining the face image of the passenger, a registration and verification request information is sent; the registration and verification request information includes face features extracted according to the face image, an SDK version number Ver used for extracting the face features, and user information of the passenger. n After obtaining the face image of the passenger, a registration and verification request information is sent; the registration and verification request information includes face features extracted according to the face image, an SDK version number Ver used for extracting the face features, S12. The gateway receives the registration verification request information and, according to the SDK version number Ver n and forwards it to the face registration server using the same SDK version number for 1:1 identity verification; S13. After identity verification, the passenger's face data is added to the feature table of the database, and the corresponding record ID is added; the server update process is executed; the face data includes user information, face features, and SDK version number Ver n ; S2. When the passenger passes through any security gate SG m m = 1, 2, …, M, M is the number of security gates, comprising: S21. Security gate SG m After obtaining the face image of the passenger, a face recognition request information is sent; the face recognition request information includes face features extracted according to the face image, the security gate SG m The SDK version number Ver used for extracting the face features m ; S22. The gateway receives the face recognition request information, and according to the SDK version number Ver m The face recognition request information is forwarded to the face recognition server using the same SDK version number for face recognition. 2.The method of claim 1, wherein, The arrangement of the security check system in the airport security check area is as follows: a plurality of independent security check channels are arranged in the airport security check area, and an entrance gate is arranged in front of each security check channel, and all the entrance gates are arranged in parallel; a plurality of entrance gates are arranged in parallel at the entrance of the airport security check area, and a passenger waiting area is formed between the queue of the entrance gates and the queue of the security check gates. 3.The method of claim 1, wherein, After the face data of the passenger is saved, the server updating process is executed, which comprises the following steps: S131. face registration server FRS i determining whether there is a face registration server FRS with a different SDK version number from the one adopted by the user i’ , and entering step S132 if there is; wherein i, i' = 1, 2, …, N ig , and i ≠ i'; N ig is the number of face registration servers S132. face registration server FRS i to the face registration server FRS i’ sending a call-off message; S133. Face Registration Server (FRS) i’ The system receives the request to extract features from the passenger's face image and obtains the data from the Face Registration Server (FRS). i’ The face data with the SDK version number is added to the feature table of the database, and the corresponding record ID is added. S134. sending an update message to all face recognition servers using the message middleware, the update message including two pieces of combined information, one piece of combined information including the SDK version number information of the face registration server FRS i and the record ID of the face data extracted for the face image in the feature table; the other piece of combined information including the SDK version number information of the face registration server FRS i and the record ID of the face data extracted for the face image in the feature table. S134. sending an update message to all face recognition servers using the message middleware, the update message including two pieces of combined information, one piece of combined information including the SDK version number information of the face registration server FRS i’ and the record ID of the face data extracted for the face image in the feature table; the other piece of combined information including the SDK version number information of the face registration server FRS i’ and the record ID of the face data extracted for the face image in the feature table. S135. Each face recognition server reads the corresponding combined information in the update message according to the SDK version number thereof.

4. A face recognition system compatible with multi-version face features, used to implement the face recognition method compatible with multi-version face features according to claim 1, characterized in that, The face recognition system comprises: The entrance gate acquisition module is configured to acquire a face image according to the entrance gate, and send registration verification request information to the face registration server after extracting face features from the face image; The gateway relocation module is configured to send the registration verification request information to the corresponding face registration server and the face recognition request information to the corresponding face recognition server according to the SDK version number; The identity verification module is configured to perform identity verification according to the received registration verification request information, and save the face data of the passenger who passes the identity verification; The server updating module is configured to detect the SDK version number of the face registration server, and acquire face data of the same passenger according to different SDK version numbers; The security check gate acquisition module is configured to acquire a face image according to the security check gate, and send face recognition request information to the face recognition server after extracting face features from the face image; The face recognition module is configured to perform face recognition according to the received face recognition request information.

5. The multi-version compatible face feature compatible face recognition system according to claim 4, wherein, The operation process of the server updating module comprises the following steps: S131. face registration server FRS i determining whether there is a face registration server FRS with a different SDK version number from the one adopted by the user i’ , and entering step S132 if there is; wherein i, i' = 1, 2, …, N ig , and i ≠ i'; N ig is the number of face registration servers S132. face registration server FRS j to the face registration server FRS i’ sending a call-off message; S133. Face Registration Server (FRS) i’ The system receives the request to extract features from the passenger's face image and obtains the data from the Face Registration Server (FRS). i’ The face data with the SDK version number is added to the feature table of the database, and the corresponding record ID is added. S134. sending an update message to all face recognition servers using the message middleware, the update message including two pieces of combined information, one piece of combined information including the SDK version number information of the face registration server FRS i and the record ID of the face data extracted for the face image in the feature table of the face registration server FRS i ; the other piece of combined information including the SDK version number information of the face registration server FRS i’ and the record ID of the face data extracted for the face image in the feature table of the face registration server FRS i’ ; S135. Each face recognition server reads the corresponding combined information in the update message according to the SDK version number thereof.

6. The multi-version compatible face feature compatible face recognition system according to claim 4, wherein, The face recognition system further comprises: The information registration module is configured to acquire a face image according to the terminal device APP, and send the face image, face features extracted from the face image, the SDK version number used by the terminal device APP to extract the face features, and user information of the passenger to the face registration server.

Images