Research On Three-frame Adaptive Threshold Algorithm For Detection And Tracking Of Moving Targets In Video Surveillance System

As an important branch of computer vision application technology,mobile target detection technology aims to achieve real-time video acquisition and detection of moving targets within the surveillance area.And it saves the information of moving targets as important evidence for analyzing field conditions to achieve the effect of active monitoring,resource saving and efficiency improving.This paper combines the advantages of the widely used inter-frame difference algorithm and background difference algorithm in target detection and tracking algorithm,and introduces a moving target detection algorithm that combines these two algorithms together.First,when the mixture Gaussian model is used to model the background,this method improves the variance update and maintains the variance for the unmatched Gaussian distribution,so that the modeling effect is more similar to the real background.Second,two frames difference is replaced by three frames difference.Adaptive difference method is adopted to obtaining the threshold.Finally,the results of these two differential processing methods are combined for further perform morphological processing and the moving target extracting.Traditional passive video surveillance systems require 24-hour video recording and manual recognition of moving objects.This article has designed an embedded intelligent video surveillance system that focuses on active monitoring.The system is mainly based on the improved moving target detection technology.In addition to realizing real-time monitoring,it is also responsible for detecting,alarming,and tracking the moving objects that enter the monitoring area.The system does not record irrelevant video data at ordinary times.When a moving object enters the monitoring area,the system will automatically detect and identify the moving target and perform audio alarm through the audio.At the same time,the system will find the video and image when the target is moving.Videos and images are saved to the system,providing an easy way for the users to access the embedded Web server through the browser.The experimental results show that,compared to the traditional algorithm,the average missed report rate of the moving target of the improved mobile target detection algorithm in this paper is increased by more than 3% at the same speed and by more than 5% under the same volume.The embedded intelligent security monitoring system not only has the functions of video acquisition,storage and network monitoring in the traditional video surveillance,but also can perform early warning,active capture and storage of the moving targets within the camera monitoring area.The system achieves the purpose of active monitoring,which has great practical value and application prospects.