Design of security system based on intelligent video surveillance (Part 2)

0 Introduction

With the continuous progress of my country's social economy, the public's awareness of security precautions is gradually improving, and video surveillance systems are increasingly being used in enterprises, schools, banks, and residential areas. However, at present, most video surveillance systems only provide real-time images to staff, which are monitored and processed manually. This monitoring method has many shortcomings: First, it is impossible for most such systems to configure cameras and monitors in a 1:1 ratio, so most monitoring points cannot be monitored all the time; second, experiments have shown that if a person continuously observes more than two monitors, he will miss 45% of the scenes after 10 minutes, and 95% of the scenes after 22 minutes, which is extremely inefficient; third, since real-time monitoring does not conform to human psychological and physiological characteristics, in order not to miss any undiscovered situation, 24-hour video recording is required for subsequent review. This generates a large amount of useless information, and a small amount of useful information is submerged in it, making it more difficult to obtain this useful information; fourth, the business function is single, and due to the lack of intelligent analysis and linkage processing capabilities, it can only "monitor" but lacks the ability to prevent and "control" disposal.

The intelligent video surveillance system uses artificial intelligence, pattern recognition, probability theory and image processing technology, and uses the powerful data processing capabilities of computers to analyze video data, filter out useless or interfering information in the image, extract key information from the video source, determine whether there are any abnormal conditions, and handle them in the fastest and best way. The intelligent video surveillance system can effectively carry out pre-warning, in-process processing, and timely evidence collection after the event. It is a fully automatic, all-weather, real-time monitoring intelligent system.

1. Functional Analysis of Intelligent Video Surveillance System

First of all, the intelligent video surveillance system can perform intelligent analysis on the collected video data and realize various applications on this basis. For example, when a target is detected entering the monitoring area, the system automatically locks the target, tracks and shoots video, and triggers the corresponding alarm linkage; it can realize face recognition and license plate recognition, and intelligently record the entry and exit of people and vehicles; combined with image fusion algorithm, it can realize the fusion of infrared and visible light images, and enhance the reliability of night monitoring.

Secondly, the intelligent video surveillance system can effectively integrate other security equipment, such as access control systems, infrared detectors, smoke and temperature sensors, alarms, etc., to achieve the integration of security work and give full play to the maximum efficiency of the system. For example, when the infrared detector finds an unknown target entering the monitoring area, the system will quickly issue a command to require the corresponding network camera to move to the corresponding preset position, sound the alarm, and prompt the administrator to check the video signal of that channel. This integrated system has a variety of monitoring methods, which can greatly reduce the false alarm rate and improve the efficiency of security work.

Third, the intelligent video surveillance system can sense the working status of the front-end camera, such as video blur, video blockage, video loss, and perspective change. For example, in a large-scale monitoring system, there are usually hundreds or even thousands of monitoring points. The on-duty personnel of the monitoring center can only monitor dozens of video images at most. When one of the videos is intentionally or unintentionally blocked, it is difficult for the on-duty personnel to detect it in time, which brings major safety hazards. The system has a certain self-detection capability, which can detect and repair system problems in time to ensure the normal operation of the system.

Fourth, the intelligent video surveillance system provides automatic evidence video retrieval for video recordings. This function uses target detection technology to obtain the type, shape, size, speed, position, color, and other specific target sign information of the target, thereby generating a rich video index and realizing specific video segment retrieval or target event retrieval. For example, if you enter the "blue hat" personnel who walked through the "6 kV power distribution room" between "August 23 14:00" and "August 23 16:00", the system can find relevant videos for users based on these key information points, greatly improving the efficiency of video analysis.

2 Architecture Design of Intelligent Video Surveillance System

2.1 System Architecture Design

The intelligent video surveillance system architecture design proposed in this paper can be divided into two parts: software and hardware (as shown in Figure 1). The software part consists of four parts: system background monitoring end, management end, service end and database; the hardware part mainly includes: network video camera, I/O electronic controller, various alarm sensors, searchlights, access control card readers, access control controllers, etc.

2.1.1 Software components

(1) Monitoring terminal: It is the information processing unit of the intelligent video monitoring system. It is responsible for connecting users with the central server, serving as an information transfer station and the key to ensuring that users can receive video information and alarms in the first place. Its main functions include: viewing real-time monitoring images, receiving alarm prompts, querying historical alarm records, and remotely controlling various alarm devices.

(2) Management side: mainly used for personnel authority configuration, hardware equipment configuration, system linkage alarm plan setting, and timing plan management.

(3) Server: As the nerve center of the intelligent video surveillance system, it undertakes many responsibilities such as hardware equipment management, data communication, historical video record management, network congestion control, and intelligent alarm processing.

In order to ensure the security, stability and ease of use of the system, the monitoring end, management end and server end are all written in C# language and run on Microsoft's WINODWS operating system.

(4) Database: Microsoft's SQL2005 database is used, which has excellent performance, relatively simple operation and high security. All hardware information, personnel information, alarm handling plans, alarm history records and video indexes in the system are stored here for easy modification and call.

3 Implementation of Intelligent Video Analysis

The acquisition and analysis of video images are mainly completed by the embedded microprocessor built into the front-end camera. This data processing method allows the system to analyze the original or closest to the original image and make a quick and accurate judgment in the first place.

A complete video image analysis and processing process requires the integration of multiple technical means such as image processing technology and pattern recognition technology to achieve better practical results. Its working process includes image preprocessing, image segmentation, feature extraction and image classification. The workflow diagram is shown in Figure 3.

The system's image recognition design is implemented by borrowing the idea of ​​motion detection: first, the background is restored based on the statistical information of the pixel value of each coordinate in the entire sequence. If there is any abnormality, it is extracted; then statistical methods are used to identify the category to which the abnormality belongs.

Image recognition is mainly achieved by using the inter-frame change detection method, and its basic process is divided into:

(1) Image preprocessing: According to the blur of the image, various special techniques are used to highlight certain details in the image and weaken or eliminate irrelevant information, so as to enhance the overall or local features of the image.

(2) Image background restoration and anomaly extraction: restore the image background based on the statistical information of the pixel values ​​at each coordinate in the entire sequence, and then subtract the current frame from the restored background to extract the area where the anomaly has occurred; (3) Image classification: subtract the current frame from the restored static background to extract all areas where anomalies may have occurred.

3.1 Image Preprocessing

Common image enhancement can be divided into two categories based on the processing scope: spatial domain and frequency domain. Spatial domain enhancement is to process directly in the space where the image is located, and directly calculate the pixel grayscale value of the image. Spatial domain image enhancement technology can be described by formula 1:

Where: F(x,y) is the image before processing, G(x,y) is the image after processing, and H(x,y) is the spatial operation function. Frequency domain image enhancement is to transform the image in the original space into another space in some form, and then use the unique properties of the transformed space to process it, and finally transform it back to the original space. The process can be described by Figure 4:

3.2 Image background restoration and anomaly extraction

There is a strong correlation between frames in a video sequence. If only a single frame is used for analysis and processing, the error rate is very high. The current method with better analysis effect is to process multiple frames together. Based on this idea, the background can be restored according to the statistical information of the pixel values ​​at each coordinate in the entire sequence. The system designed in this paper uses static background restoration for processing: