The Applied Research Of Sequence Image Tracking Algorithm Under The Condition Of Complex

Target tracking technology is widely used in military, security, trafficsurveillance. How effective tracking of the target of interest in a video sequence is avisually challenging topics.This paper studies the tracking algorithm of moving target in the imagesequence. First, studied the movement characteristics of the moving target and themodeling principles, methods of moving target; Secondly, studied of the classictracking algorithm to track moving targets in the image of the sequence, such asparticle filter algorithm, class Kalman filter algorithm, the mean shift algorithm. Thispaper presents a moving target tracking in image sequences with strong tracking filteralgorithm to track the moving target. As which the particle filter was taken into a localoptimum have to resample that increase the time overhead and Kalman filter cannotcatch up with the moving target in the case of larger changes to the environment orsuddenly appear in the case of high-speed motor based on the particle algorithm classKalman filtering algorithm, the mean shift algorithm. Strong tracking filter isimproved on the basis of the classical Kalman filter which is for the first timeproposed by Zhou Donghua in fault diagnosis. It changes the Kalman filter-likeKalman gain which is constant and stored in the computer and will not change thedefect in advance, so that the Kalman gain rapid growth in the tracked target isblocked or the sudden appearance of high-speed mobile and then normalized to zero.Strong tracking filter is to overcome this deficiency of Kalman filter, it can be betterapplied to the image sequence moving target tracking.Image sequence moving target tracking experiments show that in the case of themoving target is obscured in the role of the strong tracking filter algorithm can bevery good track moving targets, has strong robustness in the background moves, themore complex environment can also be able to fast track the objectives of the campaign, and achieved good tracking performance.The algorithm uses parallel tracking algorithm for multi-target moving targettracking, so that multi-target tracking can be achieved, but due to the time constraints,only the double-target tracking was studies. Experimental results show that thetracking effect is quite satisfactory, and track the targets which in the case of crossable to distinguish the respective tracking target and tracking moving targets. Thisdoes not appear disturbed and tracking in case of failure of the tracking target. Thathas good noise immunity and a certain practicality.