Research On Tracking Technology Of Moving Targets In Complex Background

Intelligent era is fast approaching us,including the field of human-computer interaction,auxiliary medical diagnosis,safety monitoring and military and defense continue to highlight its value.Target tracking technology is the mainstream in the field of machine vision and become a new research hotspot after the field of image processing and pattern recognition.Therefore,the target tracking technology has a very important significance in both theoretical research and practical application.Most of today's target tracking algorithms are difficult to apply in practical application scenarios.The main reason is that the target and background in the actual scene is uncertain change.The target feature model can not well distinguish the target and the background,which will lead to the failure of tracking.Tracking the fast motion target under complex background,the main solutions are that the target disturbed by the background that has similar color with target under complicated and the problem of lost target maneuvering tracking.In this paper has exploration and research on the above problems.Analyzing detailedly the traditional Camshift algorithm failures in tracking moving target in complex background,this paper improves the tracking performance of Camshift algorithm,which feature information extracted by target adaptive fusion to express target model,while the introduction of target state estimation.Firstly,in the aspect of feature fusion,the main study includes the typical color features,the edge gradient direction features and analysis of the advantages and disadvantages of each feature.The color feature and edge gradient direction features of adaptive weighted fusion based on the color differentiation of target and background from combined with characteristics of the target model,solving the problem of target vulnerable to interference the similar background color,which prepares for the follow-up tracking algorithm.Secondly,according to the traditional Cam shift algorithm for fast moving targets,especially the problems of lost tracking when target maneuvering,A method is presented for fast motion or maneuvering tracking algorithm for fast motion target based on improved kalman filter Cam shift.The algorithm determines the target motion and mobility of strength according to angle difference between the target time measurement and target orientation and corrects the predictive value of acceleration of the motion target and target prediction of state,which improves target state estimation.The algorithm will the position of state estimation of target as the next frame start search window center position and solves the problem of the traditional Cam shift algorithm lost target because target moving too fast and target displacement too more in inter frame.The simulation results show that,the method has higher tracking precision when target fast moving and maneuvering.Finally,In order to prevent the edge characteristics of target can not continue to describe the edge feature because the rotation of the target,this paper proposes dynamic update the target feature model based on edge feature matching Euclidean distance.This paper puts forward Camshift target tracking algorithm based on the multi feature fusion and motion estimation.Compared with experimental simulation,the algorithm has very high robustness for maneuvering target tracking in complex background.