Research On Target Tracking Method Based On Algorithm Fusion

With the continuous improvement of the technical level,the development of the computer,whether it is hardware or software,is changing with each passing day.The development of high-performance processors provides hardware conditions for computer vision.As an important content of computer vision tracking research,target tracking has been paid more and more attention by scholars.Target tracking technology has achieved unprecedented development.Although the target tracking technology is quite mature,it faces many challenges.In order to solve the difficulties in current target tracking,this paper proposes a model adaptive update method for improving the target,which is based on the previous research.The target model is updated after being occluded and drifting.For the traditional tracking target tracking method,it is difficult to quickly find the target after the target disappears and is obstructed.An algorithm fusion target tracking method is proposed.In order to ensure the target tracking speed and tracking accuracy,this paper proposes a target model updating method based on the relevant filtering target tracking algorithm.The related filtering target tracking algorithm can achieve high-speed tracking in tracking the target.However,the original model updating method cannot cope with the target model updating after the target occlusion or tracking failure.The performance of the tracking is greatly influenced by the target deformation and rotation.In order to improve this deficiency,in the third chapter,a new method of updating the target model based on statistical model is proposed.Because the statistical model has good invariance to the rotation and deformation of the target,it can also be used to distinguish whether the target is occluded.Experiments show that the performance of this algorithm is better than that of the original tracking method.In addition,for the target tracking problem after target occlusion and disappearance,the related filtering methods have poor tracking performance under conditions such as occlusion and disappearance of the target.In order to utilize the relevant filtering speed,it is also necessary to improve the target tracking after the target disappears.The search capability is proposed to incorporate particle filtering into the relevant filter tracking method.Since the target is not in the occlusion or disappearance state in real time,the relevant filter target tracking method is used when the target does not disappear or the drift does not occur,and the particle filter tracking method is used after the target disappears or the tracking drift occurs.Although the particle filtering target tracking method is relatively slow in tracking speed,the integrated speed is improved after the fusion particle filtering and correlation filtering methods.