Do you know a few video image information databases?

In all types of wisdom public security, Xueliang project, safe city, public security big data, police big data, public security combat program will refer to the library (public security video image information database), what is the library? What is the architecture of the view library? What is included in the view library?

1. What is the overall structure of the library?

The video image information database is the core part of video parsing. Responsible for providing data aggregation access, data processing, data resource storage, data services, data cascading, data open capabilities, view library software applications. The video image information database mainly includes the storage and management of various types of data. This layer is the core data resource of the platform, including data management, original intelligent sensing database, basic resource library and special library, which plays an important role. All kinds of databases in the video image information database are built in accordance with the data resource cataloging and data element standards issued by the Ministry of Public Security. At the same time, the video image information is managed in accordance with the requirements of the relevant specifications of the public security video image information application system. The video image information database is designed according to relevant standards such as “National Public Security Video Image Information Database Construction Networking Plan”.

2. What are the subdivisions of video image storage management?

The view library platform is divided into two categories: video image base library and video image feature library. The video image base library includes basic video library, video index library, video image information library, and human car. The control database and business application database; the video image library includes the human-car special library, the event event database, the feature database, the early-warning model library, the business special library, etc. and can be continuously expanded. The video image base library mainly stores unprocessed video images and corresponding structured basic information, structured information after parsing and association, and enters the video image special inventory storage according to business needs.

2.1 Video Image Base Library

1) Basic video library

The structured and unstructured sensing data aggregated by the view library platform needs to store a copy in the base video library without data cleaning, conversion and association fusion. The library stores the original data record as data. An important reference archive for reconstruction and data quality management. The basic video library includes various types of access data such as portrait data, video structured data, WiFi data, electric fence data, video access control data, and card-passing data. After further processing, analysis and mining, the data of the basic video library is used for the supply of each theme library.

2) Video index library

The video index library stores the network device resource information (that is, the information of the networked front-end camera, including the device ID, the home alarm, the device name, the device manufacturer, the device model, the administrative region, the longitude, the latitude, the installation address, etc.), and the original video index information. , that is, index information of historical video (including central storage and front-end storage) of all networked cameras, including storage device ID, media server ID, media type, recording start time, recording end time, media file name, bit rate, resolution Wait.

3) Video image information base

The video image data content in the basic video library is imported into the video image information database after data analysis service; at the same time, the provincial video image information analysis system focuses on personnel, vehicles, case event data and related business data of the public security business system. After the big data aggregation service, it also imports the video image information library. The storage of the video image information library can be divided into a personnel information library, a vehicle information database, and a case event library. The content of the video image information library includes the original image, the parsed person, the small car picture, and the corresponding structured information.

4) People and vehicles control database

The vehicle control database stores the control data generated by the cloth control service, including the basic information of the control and the original picture.

5) Business Application Database

The data generated by other public security system systems can be associated with video big data for collision analysis. The business application database is used to import and cache business data from other public security systems for collision with video big data, and store the associated collisions. Result data.

2.2 Thematic Resource Library

Through the association of different elements of data resources (such as people, vehicles, video, address, organization, etc.), the corresponding types of thematic resource pools are built for the application.

1) Personnel library

Personnel basic information database: permanent resident information, floating population information, key personnel information, driver information, information about the case, photo of the certificate, etc., can obtain basic information of the concerned personnel from the departmental business data resource library or the provincial level analysis system.

Pay attention to the personnel checkpoint information base: personnel pass the security check equipment of checkpoints, stations, airports, terminals, railways, etc., the photos of the personnel photographed by the human verification equipment and the time of personnel passing, etc., can pass the verification of the equipment/system Docking to get.

Associated personnel library: Personnel obtained by applying intelligent analysis algorithms from video or images, and personnel information obtained by manually marking video images through a comprehensive application platform can also be obtained from a provincial analysis system.

Other personnel information: access to Internet cafe Internet information, hotel accommodation information, station airport traffic information, etc. through docking with department-level related business systems/platforms or provincial-level related business systems.

2) Face Library

It includes resource libraries such as target face information and past person face information. It can be obtained by docking with a department-level related business system/platform or a provincial analysis system, or by applying an intelligent analysis algorithm from a video image of interest. The system database should contain three business libraries: the face capture library, the face registration library, and the blacklist library.

Face capture library: contains snapshot scene pictures, face thumbnails and structured face feature data, capture location, capture time and other information. The main business application scenario of such library is image retrieval comparison, querying the portrait of the target personnel. Information such as location, time, etc.;

Face registration library: mainly introduces some large-scale portrait images, structured facial feature data and identity information, such as a local social security portrait information database in a prefecture-level city. The main application scenario after import is image retrieval comparison. And identity information query to determine the identity of the person;

Blacklist library: contains face images of high-risk personnel, special personnel, structured face feature data and personnel identity information. The main application scenario is face comparison warning for real-time traffic flow in each face bayonet.

In general, the face capture library and the face registration library are used as static libraries, which are suitable for post-query retrieval retrieval targets and blacklist libraries as dynamic libraries for real-time comparison alarms. One or more blacklists can also be checked and controlled to form a targeted face control library, and face-to-face comparison alarms are performed with the front-end real-time video.

Among them, the flow of the library is more and more large, and the size of the storage device needs to be calculated according to the project situation. Blacklist library data is imported by public security or professionals. The storage size is generally fine-tuned, but there will be no order of magnitude change.

3) Vehicle theme library

It refers to sorting and correlating vehicle information in the basic resource pool to form a vehicle theme library.

Pay attention to the vehicle basic information database: pay attention to motor vehicle information, pay attention to non-motor vehicle information, pay attention to violation information, and pay attention to the information involved in the case. You can obtain the basic information of the vehicle from the departmental business data resource library or the provincial level analysis system.

Passing information database: including the passing information of the vehicle through the bayonet or electric police and the pictures of the photographed vehicles, mainly storing the inter-provincial bayonet and some key card-passing information, including: vehicle number plate, transit time, location, etc. information. The vehicle data and the bayonet picture can be obtained from the docking bayonet system, and the data such as the bayonet passing can be obtained from the provincial analysis system.

Model library: Vehicle character analysis and identification of vehicle pictures, forming a resource information library containing vehicle characteristics such as license plate number, vehicle brand, subdivision model, body color, etc., referring to vehicle information obtained by applying intelligent analysis algorithm from attention video or image. Or the vehicle information obtained by manually marking the video image through the integrated application platform.

4) Case event library

The data types of the case (things) feature library include video, picture, event attribute information, case related personnel, vehicle, item information, and so on.

The source of the case (things) includes all kinds of event information that participated in the research and judgment using video image data, and also from the parsed video image information in the video image information database.

The construction and analysis of the follow-up case (things) are all based on the case library.

5) Feature database

The feature data library stores the video image feature vector parsed by the video image analysis service. For different vendors of the same type of video image intelligent analysis algorithm, the analyzed feature vectors are different. Therefore, the feature data library merges the video image feature vector and the corresponding algorithm vendor identifier into the feature data library to provide data support for video image retrieval.

The feature data saved by the video image feature data library includes: pedestrian feature vector, vehicle feature vector and face feature vector, and other types of feature data can be saved according to the development of algorithms and techniques.

6) Early warning model library

The early warning model stocks the process and result data from the model analysis. The data model can be executed at the time of the service call, or automatically in the background in the form of a task, and the execution process and result storage and the topic model analysis library are used by the upper application.

7) Business special library

The thematic data generated by various business applications based on video big data (such as inspectors, public security, criminal investigation, intelligence and other specific business applications) are stored in the business special library. According to the needs of business applications, the business special library can be extracted from the basic video library. It can also be from the special library of people, cars and cases.

The accurate definition of a crusher hammer head should be the hammer head of a hammer crusher. The hammer head is one of the core components of a hammer crusher, arranged on the hammer shaft of the crusher rotor. The hammer head directly strikes the material during high-speed operation of the crusher, ultimately crushing it into a suitable material particle size. The hammer heads of crushers on the market can be divided into two types based on manufacturing processes: casting and forging, but their wear resistance is different. Due to the need for good wear resistance at the head and sufficient toughness at the handle when crushing materials, materials such as alloy steel, high manganese steel, cast steel, and high chromium cast iron bimetallic composites are usually formed in one forging or casting process, with corresponding heat treatment processes being more economical. Crusher hammers can be divided into three types based on material: high manganese steel hammers, bimetallic hammers, composite hammers, large gold tooth hammers, and medium chromium alloy hammers, The main factor is the content of silicon dioxide, and modern industrial technology shows that using hard alloy hammers performs well in terms of wear resistance.