Research On Intelligent Monitoring Technology Of Moving Pollutants Based On Background Difference And Adaptive Tracking

Moving target tracking is an important research field of computer vision technology.Due to the scale change,the target being occluded in a large area,and the high similarity between the target and the background,the problems of target recognition and moving target tracking have brought many research difficulties.Multi-frame difference method,mean shift and motion estimation provide reliability for the above difficulties solution.This thesis takes the detection and tracking of moving pollutants as the application target.First,a method for detecting moving pollutants based on the mean background method is designed,and then a tracking algorithm for moving pollutants based on state estimation adaptive drift is designed to solve the problem.A problem that is difficult to track when the target pollutant is blocked.The innovative work of this thesis is as follows:(1)Aiming at the situation that the traditional average background method is easy to deform in the dynamic environment,the background model update method under the dynamic background is proposed.The pre-background model and the background incentive mechanism are added to effectively overcome the failure problem and residual image of the background model.Problem,optimize the efficiency of background segmentation before and after.(2)Aiming at the situation that the traditional difference method will produce residual images and target holes,an average background subtraction method based on five-frame difference is proposed.By performing differential operations on five consecutive frames of images,the contour information extraction of the foreground target can be effectively guaranteed,and the double shadow problem of the detected target can be effectively solved at the same time.In the comparison experiment of different video data,this method can effectively reduce the extraction error of the foreground target,and the contour information obtained by segmentation is complete and clear.(3)The motion pollutant tracking algorithm based on the adaptive drift of state estimation continuously tracks the motion pollutants,and adopts an adaptive search window to deal with the tracking loss caused by the target scale transformation.When the moving target is not occluded,the feature migration algorithm based on the chromaticity histogram model is used to obtain the target offset;when the moving target is occluded,the adaptive drift tracking algorithm based on state estimation is used to display the abnormal situation of the moving target Perform motion estimation and prediction.In different video data comparison experiments,it can effectively solve the tracking loss problem caused by the disappearance of the target.The experimental results show that the method can effectively detect and track the moving pollutants.