Research On Tracking Algorithm Of Infrared Ground Target

Nowadays,with the continues development of high-tech technology,ground target tracking and recognition technology in the visible field has been unable to meet the needs in the fields of national defense and military and civilian monitoring.Especially for the night of the ground target tracking,it is difficult for the target tracking in the visible light.Infrared imaging technology can make a clear image display in all days,and it also has a good anti-jamming capability in some harsh weather,so it has been rapid development in the field of precision guidance of the military,and it is one of the main research direction in precise guidance against ground targets.At present,there are some shortcomings exist in the mainstream infrared tracking algorithm,such as target tracking drift,target occlusion and bad real-time problem.Based on this,in this paper,a series of perfections and improments are made on the basis of the research and discussion of several common target tracking techniques.And the main work of this thesis is as follows:(1)This thesis introduces the research background of infrared ground tracking technology and its development of domestic and foreign,and describes the research significance of infrared ground target tracking.This paper introduces several commonly used infrared ground tracking technology theory and implementation steps,and discuss and analyze the advantages and disadvantages of these algorithms.(2)In this thesis,because the texture and grayscale features of the feature pool extracted by the compressive tracking algorithm are not uniform,a new random measurement matrix is constructed.It effectively obtain the target characteristics,and a variety of features are weighted to construct the target model,which improves the reliability of algorithm tracking performance.(3)Aiming at the situation that the target tracking box can't change adaptability with the size of the infrared target in the tracking process of the infrared ground target,an improved compressive tracking algorithm of infrared ground target scale adaptability is constructed.In the case of the infrared ground target occlusion,the Kalman filter algorithm is used to predict the tracking position of the infrared target,which solves the problem of tracking in the case of the infrared ground target occlusion with short time.In this thesis,through the experimental test and comparative analysis,it is proved that the improved algorithm(Improved Compressive Tracking,ICT)improves the tracking stability in the infrared target tracking process.(4)Aiming at the problem that when the infrared target changes too much,the tarcking effect of ICT algorithm is reduced,a infrared target tarcking algorithm based on TLD(Tarcking-Learning-Detection)algorithm and ICT algorithm is researched.And the three-frame difference algorithm id added to the TLD detector module to optimize the search area of target,which improves the tracking speed and tracking accuracy of the algorithm.Through experimental tests,the results of the experimental tests show that the algorithm of this thesis effectively improves the infrared ground tracking effect,and it has good accuracy and real-time.