Analysis of Distributed Intelligent Video System

Intelligent video analysis technology needs to have three levels:

Artificial intelligence-based technology can separate the target object in the foreground from the background;

· Data that independently extracts details of the foreground object, i.e. video Meta Data;

These Meta Data can also be used to separate some events by setting certain rules and provide them to system integrators in the market.

There are currently four common architectures for where intelligent video should be set up in a surveillance system:

· Embedded application. Make full use of the abundant computing power of network cameras to realize intelligent video;

· Multi-chip type. Use additional dedicated CPU or DSP to realize intelligent video;

· Through application software running on DVR or PC;

· Realize intelligent video surveillance by leveraging front-end embedded applications and back-end software functions.

Distributed intelligent video task processing

The intelligent video network monitoring system based on the distributed enhanced image processing structure platform can automatically identify a variety of sensitive events through its intelligent video analysis function, thereby providing a higher security level and management for the user's security system (as shown in the figure).

The distributed enhanced image processing structure divides the traditional processing tasks into two independent parts. The camera performs the front-end processing, while the surveillance recorder or management software is responsible for the back-end processing. The intelligent network camera in the front-end processing part first separates the target from the environmental interference, then extracts the moving or fixed target, packages the target information and sends it as a separate data stream. The back-end processing is completed by the intelligent network surveillance recorder or RSM intelligent monitoring management software. The back-end receives and records the target data sent from the camera, then screens the targets or events that meet the preset conditions, and finally generates an alarm and performs associated recording.

Advantages of Distributed Intelligent Video System

The advantages of distributed intelligent video systems can be summarized as follows:

1. Front-end processing:

· Users can achieve efficient intelligent video analysis with relatively little investment. Its distributed structure distributes the analysis tasks to the front-end camera and the back-end software or NSR series network surveillance recorder, so that the entire system has efficient video analysis capabilities, while the back-end does not have heavy processing tasks, and users do not need to invest in large amounts of hardware;

Enhanced motion detection processing effectively prevents false alarms caused by environmental factors. By comparing the current image with the previous 15 frames, this technology can reduce interference caused by environmental factors, such as the slight movement of leaves or water, which other systems often generate false alarms due to environmental interference;

The front-end intelligent network camera extracts the target from the original uncompressed image, so the intelligent video analysis has higher accuracy. The traditional intelligent analysis system analyzes and processes the compressed video. Video compression will lose some real information and generate some noise signals, which will result in more missed alarms and false alarms.

2. Back-end processing: The back-end only analyzes the target data information sent by the front-end camera, without processing the original video signal, so the back-end system can complete efficient intelligent video analysis without expensive equipment. The associated screen display of the alarm event enables the monitoring personnel to easily complete the monitoring of the entire system; and the event-related recording function makes the video search and playback simple and fast.

3. Network transmission: The intelligent monitoring system based on distributed enhanced image processing structure technology realizes real on-demand monitoring. Only when an alarm occurs, the relevant video is sent to the back-end for monitoring and recording. In general, only a small amount of target data information needs to be transmitted through the network, and the flow of these target data information is less than 1/50 of the total video volume, thus greatly reducing the requirements and consumption of network bandwidth.

Application and examples of distributed intelligent video system

Distributed intelligent video systems have a wide range of applications and can provide many functions for research and development and implementation. In summary, these application functions can generally be divided into two major areas.

The first is security-related applications, which are also the most important applications of intelligent video analysis. Its main function is to assist security departments in improving the security protection of important parts and important items, and identifying suspicious persons. Such applications mainly include: advanced video motion detection, object loss or displacement detection, object tracking, facial recognition, vehicle recognition, illegal detention, fireworks detection, etc.

The second is non-security related applications. They are mainly used in retail, service, management and other industries as auxiliary tools for management and service to improve service levels and turnover. Such applications mainly include: people flow statistics, crowd control, behavior analysis, attention control, traffic flow control, etc.

The application of distributed intelligent video systems in the construction of digital "safe city" projects is a particular concern of today's society. It includes intelligent monitoring such as no parking, one-way driving, people counting, valuables locking, access control, no climbing, exit monitoring, people detection and warning areas.

The application solution integrates a camera and an embedded video analysis module on a circuit board to form a compact and low-cost high-tech product. After setting the sensitive detection area, it can automatically detect, track and count the flow of people entering and leaving the camera, and store accurate statistical data in the database, which can also be retrieved from the database using browsing software.

Crowd counting can provide business analysis information for the retail industry, helping them understand the shopping habits and advertising effectiveness of retail stores, so as to effectively allocate and manage sales or service personnel. By comparing customer transaction volume and point of sale information, some important data can be obtained. Public places can use this information to estimate the number of spectators for safer control, such as stadium crowd counting, which can monitor the number of people in the stadium. In the monitoring screen of the panoramic surveillance camera, when the number of people exceeds the preset value, the on-duty personnel of the monitoring center can be prompted to check.

The solution can also be combined with supporting access control equipment, perimeter protection equipment and site protection equipment to form a comprehensive intelligent video analysis solution. Using proprietary computer vision technology, it can achieve two-way tracking and counting of multiple people/vehicles even in high-density environments.

The application solution is also equipped with corresponding application software, i.e. a web-based application (such as iSense Reporter), which can easily retrieve statistical data stored in the database and centrally access the data of interest by selecting cameras and time. Three different report types and three different report graphs can support the generation of nine report forms for customer reference.

Several issues that need attention in the plan

It should be an integrated design: a compact and powerful device with a built-in camera, which saves equipment costs and simplifies installation and maintenance. As long as the distributed intelligent video system is connected to the network, it can be configured and operated;

It should have video technology based on unique technology, efficient image processing and parallel computing to achieve detection, tracking and accurate counting of targets. It should be able to handle interference in different scenes and environments with multiple targets, and achieve high accuracy with minimal setup work;

· Simple to use: Easy to install and operate. Based on advanced recognition algorithms for scene objects, operators can set specific sensitive areas. For people counting, the size of people can be automatically calibrated during the recognition stage.

In addition to the above-mentioned application of distributed intelligent video systems in personnel counting solutions, DVRs with intelligent video analysis functions can also be developed according to market demand. Intelligent DVRs embed advanced video intelligent analysis technology into hard disk recorders and combine it with monitoring technology, which not only greatly enhances the effectiveness of video monitoring systems, but also can detect crises and respond in a timely manner at the beginning of potential dangers.

The DVR uses a high-performance DSP processor to process, analyze, filter, and identify surveillance image data at high speed, extract key information from it, and includes image recording, security rule comparison, alarm control management, network video transmission, centralized control management and other functions. It is a complete DVR system with intelligent video analysis function. It can realize high-definition D1 format hard disk recording, which is suitable for ATM machine surveillance recording applications. The intelligent DVR has five optional intelligent video analysis functions such as leftover detection, personnel counting, trip wire alarm, area alarm, and abnormal face detection.