A design method for distributed face basic information intermediate library service compatible with multiple algorithms
By designing a distributed face basic information intermediate library service that is compatible with multiple algorithms, the data security and privacy protection problems of face recognition system are solved, data separation and secure query are realized, high concurrency and large data management needs are met, and security is provided. Reliable distributed services.
Patent Information
- Application Number
- CN201910399714.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-05-14
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2039-05-14
AI Technical Summary
The existing facial recognition system has defects in data security and privacy protection. The facial feature data of various algorithm manufacturers cannot be universal, and there is a lack of effective privacy protection and update mechanisms, resulting in high risk of privacy leakage and a lack of mature distributed facial foundation. Information intermediate library solution.
A distributed facial basic information intermediate library service that is compatible with multiple algorithms is designed. By creating a personnel basic information database, photo database and feature value database, the three libraries are separated, and data changes are monitored and processed in real time, providing feature extraction and encryption. The mechanism combines the secure API interface for permission control and log traceability to ensure data security.
It realizes the separation and query services for facial recognition applications while ensuring data security, supports permission control of different types of personnel information, flexibly verify the identity and IP access of algorithm manufacturers, and meets the management of large data volume and high concurrency requests. Quickly return personnel information and pictures.
Smart Images

Figure CN110334225B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data information intermediate library security technology, and in particular to a related design method for a distributed human face basic information intermediate library service that is compatible with multiple algorithms. Background Art
[0002] At present, facial recognition application systems at home and abroad independently maintain information, photos and facial feature libraries of various types of personnel. The facial feature extraction algorithms, data structures, storage methods and comparison algorithms of facial recognition algorithm manufacturers are different. The facial feature data of different algorithm manufacturers cannot be used interchangeably. Basic facial information and personnel photo data lack a complete privacy protection and update mechanism. When using the facial recognition algorithm system, it is necessary to first enter or batch import personnel information and photos into the facial recognition system, establish a personnel information library, use facial photos to extract facial features, establish a corresponding facial feature library, and then query personnel information and original photos from the personnel library of each system based on the input facial photos and feature value library. The facial recognition algorithm system saves complete personnel information and photos, which poses a huge risk of privacy leakage;
[0003] The research on the basic face information intermediate database technology is still in the exploratory stage at home and abroad. There is no replicable case, which is a challenging topic.
[0004] Basic facial information includes basic information and photos of various personnel. We build a distributed basic facial information database and related services, referred to as the "intermediate database". The intermediate database is built at the top-level business center, and sub-nodes are built in various business units to establish a distributed data service system to provide basic data services for the face recognition application system;
[0005] In order to ensure the security of personnel information and photos, prevent information abuse and protect citizens' privacy, the intermediate library service needs to provide a complete set of mechanisms including person-photo separation, information encryption, feature extraction scheduling and distribution, and information security query. Provide distributed services for basic facial features and personnel information query for applications such as identity verification, portrait checkpoints, pedestrian red light violation capture, and police WeChat;
[0006] The basic facial information intermediate library needs to separate personnel information and photos, encrypt sensitive information, and cooperate with various facial recognition algorithm manufacturers to build a facial feature extraction system. The intermediate library will dispatch various systems to extract facial features, process incremental data, and synchronously distribute feature data. It also provides a secure API interface service for facial recognition applications to request basic facial information, and needs to provide permission control, sensitive information filtering, request limit management, image watermarking, and log tracing functions. In summary, there is no mature solution for the basic facial information intermediate library service in China or even around the world. Summary of the invention
[0007] In view of the problems existing in the above-mentioned prior art, the present invention proposes a distributed human face basic information intermediate library service compatible with multiple algorithms, fully considers various possible security risks, realizes the separation of three libraries, and provides safe and reliable distributed services for various application systems;
[0008] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:
[0009] The present invention provides a design method for a distributed human face basic information intermediate library service compatible with multiple algorithms, which comprises the following steps:
[0010] S11: Create a personnel basic information database, a personnel photo database and a facial feature value database. The personnel basic information database is used to store structured data of personnel basic information, the personnel photo database is used to store personnel photos, and the facial feature value database is used to store facial feature value data. The facial basic data includes personnel basic information and photos. The personnel basic information database includes a personnel type data table (regular personnel, foreign personnel, fugitives);
[0011] S12: Real-time monitoring of changes in basic information data of personnel, including addition, modification and deletion of basic information data of personnel. When adding or modifying basic information data of personnel, the feature management service will push the photo corresponding to the basic information of the personnel to the face feature extraction system for face feature extraction, and send the extracted face feature value to the face recognition algorithm manufacturer. When deleting a photo, it is necessary to add a deletion feature mark in the face feature value database;
[0012] S13: When the facial feature extraction system returns data, the feature management scheduling service calls a feature callback interface to receive the facial feature information pushed by the facial feature extraction system in real time, and stores the received facial feature data in the feature value intermediate library;
[0013] S14: The multi-algorithm vendor face recognition system obtains face feature value information from the feature value intermediate database, creates different local portrait libraries according to the corresponding personnel types in the personnel type data table, and the name of the portrait library is the name of the personnel type. Then, the face feature value information is sent to the corresponding portrait library for application.
[0014] S15: After the algorithm manufacturer obtains the globally unique number of the person by comparing the captured photo with the facial feature value library, the algorithm manufacturer can call the security API interface to obtain the detailed information and photo information of the person;
[0015] Preferably, the step S12 further comprises:
[0016] S121: real-time monitoring of data changes in the personnel information database, including personnel addition, deletion and information update;
[0017] S122: Pushing the changed face photo data to the feature extraction system of the algorithm manufacturer to perform face feature value extraction operation;
[0018] Preferably, the step S13 further comprises:
[0019] S131: receiving the extracted facial feature information pushed by the feature extraction system in real time, and performing data verification and validation functions;
[0020] S132: The personnel feature data received by the feature management scheduling service is stored in the feature value intermediate library, and the feature extraction task status is updated, including normal, face detection failure, image download failure and other error status;
[0021] Preferably, the step S14 further comprises:
[0022] S141: The face recognition system monitors the face feature value database in real time. When a new face feature is found, the face feature data is synchronized to its own system in real time. The status of the face feature value data includes addition, modification, and deletion.
[0023] Preferably, the step S15 further comprises:
[0024] S151: To obtain detailed information and photos of a person, the face recognition system first needs to call the authentication interface of the security API service and pass in the application identifier and application key provided in advance for authentication;
[0025] S152: Get the token value returned after authentication, and pass it into the globally unique number and related parameters of the person's feature data obtained by the face recognition system;
[0026] S153: After entering the secure API interface service, the server will verify the caller's authority, record logs, and add photo watermarks, a series of security operations;
[0027] S154: Finally, detailed basic information and photo information of the personnel are returned;
[0028] The present invention also provides a distributed human face basic information intermediate library service compatible with multiple algorithms, characterized in that it includes: a personnel basic information library (information library + photo library), a feature extraction management service and a security API service, wherein:
[0029] The personnel basic information database is used to store personnel basic information and photo information;
[0030] The feature extraction management service is used to schedule various algorithm vendors to perform personnel feature extraction management. The intermediate library feature extraction scheduling service sends feature extraction requests to each algorithm feature extraction system. At the same time, the intermediate library feature extraction scheduling service provides a callback interface for returning facial feature data.
[0031] The security API service is used to display photos after matching the characteristics of people. It is recommended to use the interface for obtaining personnel information in batches to obtain photos and basic information of the matched people, and then call the interface for obtaining detailed personnel information to obtain the information and original photos of a single person according to the needs of the user operation on the interface;
[0032] Preferably, the secure API service includes:
[0033] Security API authentication interface, permission control, sensitive information filtering, request quota management, image watermarking, and log tracing;
[0034] Compared with the prior art, the present invention has the following advantages:
[0035] (1) The distributed basic face information intermediate library service provided by the present invention is compatible with multiple algorithms, which realizes the separation of basic information of personnel and face feature data, and provides services for various face recognition applications while ensuring data security;
[0036] (2) Support query permission control for different types of personnel information requested by applications of each branch and business unit;
[0037] (3) Ability to flexibly verify algorithm manufacturer’s identity information, IP access control, and daily interface call count;
[0038] (4) Distributed architecture, which can meet the needs of large data volume management and high concurrent data requests, and quickly return personnel information and images;
[0039] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings:
[0041] Figure 1 A diagram of the operation steps of the distributed face basic information intermediate library service compatible with multiple algorithms of the present invention;
[0042] Figure 2 This is a flow chart of feature update extraction of embodiment 2 of the present invention;
[0043] Figure 3 This is a flowchart of feature data callback of embodiment 2 of the present invention;
[0044] Figure 4 This is a diagram of the overall operation steps of the face basic information intermediate library feature management service of the present invention;
[0045] Figure 5 This is a flow chart of sending feature value data to a face system according to Embodiment 2 of the present invention;
[0046] Figure 6 Operational flow chart of the security API interface service call of embodiment 3 of the present invention
[0047] Figure 7 This is a diagram of the operation steps of the complete process of the secure API interface of the basic face information intermediate library of the present invention; DETAILED DESCRIPTION
[0048] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given. However, the protection scope of the present invention is not limited to the following embodiment; the flowchart of the design method of the distributed face basic information intermediate library service compatible with multiple algorithms of the present invention is as follows: Figure 1 As shown, it includes the following steps:
[0049] S11: Create a personnel basic information database, a personnel photo database, and a facial feature value database, storing data including personnel structured data and photos and facial feature values;
[0050] S12: When the basic facial information data in the intermediate database is updated, each algorithm feature extraction system can be directly scheduled to perform facial feature extraction, thereby achieving the purpose of automatically extracting facial feature data;
[0051] S13: when data is returned from the feature extraction system, a feature callback interface service is called to store the received feature value into a face feature value library;
[0052] S14: The face recognition system of multiple algorithm vendors obtains feature value information from the feature value intermediate database;
[0053] S15: After the algorithm manufacturer compares the original face photo features obtained by the face recognition system with the feature library, the security API interface can be called according to the global personnel number of the compared face features to obtain the detailed information and photo information of the person;
[0054] The following describes steps S11 to S15 in detail in conjunction with a specific embodiment;
[0055] Embodiment 1:
[0056] This embodiment describes step S11 in detail, which specifically includes the following steps:
[0057] S111: Create a personnel information database, a personnel photo database and a facial feature value database. The personnel information database is used to store the structured basic information of personnel, including the personnel's global unique number, name, ID card, gender, personnel photo path, and personnel tag. The personnel photo database is used to store the binary data of personnel photos, which is stored on the image server. The facial feature value database is used to store the facial feature data extracted by the feature extraction system.
[0058] S112: Importing basic personnel information data and photos into corresponding databases and image servers respectively;
[0059] Among them, step S111 specifically needs to create a face information database and related table structures (MYSQL database), and the photo database is stored using a file server, and the corresponding photo URL address will be stored in the corresponding face information structured database;
[0060] Embodiment 2:
[0061] Combination Figure 2 , 3, 5, this embodiment describes in detail another method of S12, S13, S14, which includes the following steps:
[0062] S121: real-time monitoring of data changes in the personnel information database, including changes in personnel structured data information and changes in personnel photos;
[0063] S122: Pushing the changed personnel data (here, the URL address of the personnel photo) to the feature extraction system for feature extraction operation;
[0064] S131: receiving the personnel feature information pushed by the feature extraction system through the callback interface service in real time, and verifying the data; S132: storing the received personnel feature data in the feature value intermediate library, and updating the feature extraction task status, including normal, face detection failure, image download failure and other errors;
[0065] S141: After the feature callback service interface is operated, the received facial feature data is stored in the facial feature value database;
[0066] S142: The database access method is provided to face algorithm manufacturers so that the algorithm manufacturers can detect changes in feature data and perform real-time synchronization operations;
[0067] Combination Figure 4The feature value management process of the basic face information intermediate library of the present invention is described in detail, which includes feature management service, feature callback service, and face recognition manufacturer. The feature management service will monitor the changes of personnel information in the face information database. If there are changes, the new face data will be pushed to the feature extraction system of the algorithm manufacturer. The feature extraction system will extract features from the personnel photos and call back the feature callback service interface. If the extraction is successful, the feature callback service will store the feature value in the face feature library. The algorithm manufacturer accesses the user to read the feature value data according to the provided feature value database and downloads it to the local system for face recognition comparison.
[0068] Embodiment 3:
[0069] Combination Figure 6 This embodiment describes in detail another method of S15, which includes the following steps:
[0070] S151: To obtain detailed information and photos of a person, the face recognition system first needs to call the authentication interface of the security API service and pass in the application identifier and application key provided in advance for authentication;
[0071] S152: Get the token value returned after authentication, and pass it into the globally unique number and related parameters of the person's feature data obtained by the face recognition system;
[0072] S153: After entering the secure API interface service, the server will verify the caller's authority, record logs, and add photo watermarks, a series of security operations;
[0073] S154: Finally, detailed basic information and photo information of the personnel are returned;
[0074] Combination Figure 7 , if each facial recognition manufacturer wants to obtain detailed information and photos of a person, it must be implemented through the secure API interface service. The secure API interface provides authentication, heartbeat keep-alive interface, and interface for obtaining detailed information of a person. The service authenticates based on the application identification information passed in by the facial recognition system, and returns a token value upon success. The facial recognition system will verify the authority and various security mechanisms before using this token value to call the interface for obtaining detailed information of a person. Finally, the requested record will be printed and stored in the log, and the returned photos will also be watermarked to ensure security. Disclosed herein are only preferred embodiments of the present invention. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, and is not intended to be a limitation of the present invention. Any modifications and changes made by any person skilled in the art within the scope of the specification should fall within the scope of protection of the present invention.
Claims
1. A design method for a distributed face basic information intermediate library service compatible with multiple algorithms, characterized in that: The following steps are involved: S11: creating a personnel basic information database, a personnel photo database and a facial feature value database, wherein the personnel basic information database is used to store structured data of personnel basic information, the personnel photo database is used to store personnel photos, and the facial feature value database is used to store facial feature value data; Basic facial data includes basic information and photos of people; The basic personnel information database also includes personnel type data tables including regular personnel, foreign personnel, and fugitives; S12: Real-time monitoring of changes in basic information data of personnel, including addition, modification and deletion of basic information data of personnel. When adding or modifying basic information data of personnel, the feature management service will push the photo corresponding to the basic information of the personnel to the face feature extraction system for face feature extraction, and send the extracted face feature value to the face recognition algorithm manufacturer. When deleting a photo, it is necessary to add a deletion feature mark in the face feature value database; S13: When the facial feature extraction system returns data, the feature callback interface service is called to receive the facial feature information pushed by the facial feature extraction system in real time, and the received facial feature data is stored in the facial feature value database; S14: The multi-algorithm vendor face recognition system obtains face feature value information from the feature value intermediate database, creates different local portrait libraries according to the corresponding personnel types in the personnel type data table, and the name of the portrait library is the name of the personnel type. Then, the face feature value information is sent to the corresponding portrait library for application. S15: After comparing the captured photo with the facial feature value library, the algorithm manufacturer can obtain the person's global unique number, and then call the security API interface based on the person's global number to obtain the person's detailed information and photo information; The step S15 further comprises: S151: The face recognition system calls the authentication interface of the security API service; S152: Inputting into the face recognition system a globally unique number of the person corresponding to the facial features in the captured photo and the facial feature value library; S153: The security API interface service verifies the caller's authority, records logs, and adds photo watermark security operations; S154: Return detailed basic information and photo information of personnel; The step S151 further includes: the face recognition system calls the authentication interface of the security API interface service and needs to pass in the assigned application identifier and application key. After verification, a token value will be returned, and then the token value can be used to request to obtain the basic information and photo information of the personnel. In order to maintain the validity of the token value, the heartbeat keep-alive interface of the security API service must also be called; the step further includes: a basic face information library, a feature extraction management service, an intermediate library feature extraction scheduling service, and a security API interface service, wherein: The basic face information database is used to store basic information and photo information of personnel; The feature extraction management service is used to schedule various algorithm vendors to perform personnel feature extraction management; The intermediate library feature extraction scheduling service sends feature extraction requests to each algorithm feature extraction system and provides a callback interface for returning facial feature data; The secure API interface service is used when the algorithm manufacturer needs to display basic information and photos of people after comparing the acquired original photo feature values with the personnel features in the facial feature value database. It is recommended to use the interface for batch acquisition of personnel information to obtain photos and basic information of the matched personnel, and then call the interface for obtaining detailed personnel information to obtain the information and original photos of a single person according to the needs of the user's operation on the interface.
2. According to the design method of a distributed face basic information intermediate library service compatible with multiple algorithms as described in claim 1, it is characterized in that: The step S12 further includes: the face feature extraction system of the algorithm manufacturer must be directly connected to the feature management service, forming a small local area network inside to ensure the security of data information and prevent the photos pushed to the face feature extraction system from being leaked; At the same time, the feature management service will monitor the facial feature extraction system, including network monitoring, disk reading and writing, and image data transfer, and support facial feature extraction with a concurrent volume of no less than 200 images per second.
