Research On High-altitude Projectile Detection Technology Based On Machine Visio

Machine vision detection technology has the characteristics of non-contact,high efficiency and strong practicability,and is widely used in security systems,traffic monitoring,intelligent detection and industrial automation.The potential harm of high-altitude parabolic objects to public safety is extremely serious,and it is a social problem with high attention.At present,the detection method of high-altitude parabola has problems such as passive detection,poor environmental adaptation and poor real-time performance.Therefore,it is of great significance to conduct in-depth research on high-altitude parabolic detection technology.The detection of high-altitude parabola mainly includes the detection and tracking of moving objects.This thesis firstly analyzes and compares the moving target detection algorithms such as optical flow method,frame difference method and background difference method,and focuses on the mixed Gaussian background modeling algorithm in the background difference method.In order to reduce the influence of illumination on the background difference method,the mixed Gaussian background modeling algorithm is improved by combining the edge detection three-frame difference method,and good simulation results are obtained.In terms of moving target tracking,the Camshift target tracking algorithm is analyzed.In order to solve the problem of tracking failure when the target is occluded,the Kalman filter is combined with the Camshift target tracking algorithm.Finally,the Hungarian algorithm is integrated to achieve the purpose of being able to track multiple targets stably and simultaneously,and its reliability and effectiveness are verified by experiments.In this thesis,a camera and NVIDIA Jetson TX2 embedded GPU platform are used to build a high-altitude parabola detection system.The mixed Gaussian background modeling algorithm and Kalman filter tracking algorithm researched and improved in this paper are used to detect multiple sets of high-altitude parabolas.The results of different object detection and related The data is analyzed,and the average recognition accuracy rate can reach more than 90%,and the average detection time is 8.3ms.Experiments show that the detection technology studied in this paper has achieved good results in high-altitude parabolic detection,and the stability and realtime performance meet the requirements.