Research On Target Tracking System Based On Correlation Filter Algorithm

Target tracking is one of the key research fields in machine vision.With the development of computer hardware and algorithm theory,researchers at home and abroad are paid more and more attention to target tracking in the field of national defense and civil safety monitoring.This paper mainly studies the related problems of the tracking algorithm of the moving target in the video sequence,analyzes the principle of the target locating machine,proposes a scale-adaptive target tracking algorithm based on the correlation filter and the effectiveness of the algorithm is verified through experiments.The mechanical structure of the tracking platform is designed and improved.The main contents of this paper are as follows:Firstly,the research made a conclusion of current situation of target tracking at home and abroad.By introducing and comparing the current popular target tracking algorithms,the KCF algorithm is determined as the basic algorithm in this paper.And based on this,the scale adaptive filtering algorithm based on feature fusion is improved.Secondly,established the localization model of the target tracking system.The model of the target location equation is analyzed and the exact location equation of the target in the geodetic coordinate system is obtained by using the homogeneous transformation.Based on the model,the influencing factors of the positioning accuracy are analyzed,and the main factors affecting the positioning accuracy are obtained.Then,starting from the overall system design,the overall system design is proposed.To track the prototype system,the design and improvement of tracking the mechanical structure of the head,picked a suitable ultrasonic motor,webcam and other components.And the addition of accessories such as conductive slip rings to further improve the tracking system.Finally,the algorithm proposed in this paper is simulated on the computer platform.By tracking the color video sequences in the open tracking database with different traits and comparing with the current popular tracking algorithms.The proposed algorithm achieved the best tracking results and the real-time tracking speed of 76 frames / s.And in the real environment UAV running on the state of tracking experiments,it also achieved good tracking results.