Research On Moving Target Detection And Tracking Method Of Security Patrol Robot

The intelligent video surveillance system uses computer vision and image processing technology to actively analyze the video captured by the camera in real time,completes advanced video image processing such as moving target detection,tracking and behavior analysis in the video image,realizing real-time monitoring of public safety in the monitoring area.The security patrol robot is equipped with monitoring equipment,which realizes monitoring under the state of motion.Compared with the traditional monitoring system,the camera is fixed and moves from passive monitoring to active monitoring to achieve a wider range.Moving target detection and tracking is the key technology of intelligent video surveillance system,which is widely used in military,transportation,industrial,biomedical and security.After research and development,it has made considerable achievements,but there is still no algorithm that can be applied to any scene,and it still faces challenges.This paper relies on these results to study the detection and tracking of moving targets in the mobile background.The specific work is summarized as follows:Firstly,it summarizes the research status and significance of the intelligent video surveillance system in foreign countries and China,discusses the difficulties and problems of the detection and tracking of its key technical moving targets,and draws the subject of this paper: the detection and tracking of the moving target of security patrol robots.For targeted analysis in subsequent chapters.Secondly,in the aspect of moving target detection,the advantages and disadvantages of several commonly used detection algorithms under static background are studied and analyzed,and relevant experiments are compared.The background of the surveillance image follows the movement of the patrol robot.For the shortcoming of the static background algorithm,the traditional global motion compensationalgorithm based on ORB feature point matching is used to complete the detection of the moving target.To determine the target category and whether it is a specific tracking object.In this paper,the Faster R-CNN deep learning target recognition algorithm is studied to identify the moving targets detected in the previous period and prepare for the subsequent specified target tracking.Then,in terms of target tracking algorithm,the implementation process and tracking effect of CamShift algorithm are analyzed and studied.Aiming at the problem that the target in the mobile environment is poorly tracked or even lost due to external environment such as occlusion,illumination changes and color similarity,this paper proposes an improved target tracking algorithm based on Kalman filter and CamShift algorithm.The moving target is disturbed,and the Kalman predictive value is used instead of the CamShift optimal position calculation value,and the new Kalman filter observation value is used to effectively overcome the above problem.Finally,the experimental platform is constructed,and the moving target detection and tracking algorithm experiments are performed on the acquired video images on the PC.The experimental results show that the global motion compensation algorithm can detect the moving target from the background image and prepare for the target tracking in the moving background.The CamShift algorithm based on Kalman filtering solves the above problems to some extent.Target tracking has good robustness and good tracking performance.