Research On Target Detection And Tracking Based On Background Motion Compensation

Target detection and tracking technology under the background of dynamic for very high practicability but has great difficulty,so it has been a hot issue,and can be used in many ways,such as mobile phone tracking shot,the UAV aerial,precisely guided weapons etc..The traditional detection algorithm is fast,but it is difficult to detect the target accurately in the dynamic scene.In order to detect and track the moving target quickly and accurately in the dynamic scene,a method of target detection and tracking based on background motion compensation is proposed in this paper.Firstly,the feature points of the image are extracted from the two consecutive video sequences,and the parameters are estimated by using the feature points.In order to get the motion parameters of the camera accurately,we need to filter these feature points and remove the feature points on the moving targets.Get an accurate model of the camera can be carried out after background compensation by difference algorithm of continuous image between the two frames subtraction,which can detect the target.Continuous tracking is adopted based on mean shift tracking method.The results will be used as the initial value of tracking.While in the tracking process,the algorithm monitors the similarity between the candidate template and the target template.When it drops to a certain extent,enable the detection algorithm again to relocation of the target,which can avoid the error of iterative superposition.Therefore,the tracking results are more accurate.In this paper,the corresponding experiments of target detection and tracking are presented.At the end of this paper,combined with the experimental results,the method is analyzed.