Research On Ground Moving Target Tracking Algorithm Of Unmanned Aerial Vehicle

In many UAV applications,the detection and tracking of targets are included.Therefore,the detection and tracking of ground moving targets of drones has great research value and wide application requirements.At present,most studies assume that the background environment is very ideal.In reality,not only the environment is complex,but also the distance of the target from the camera will affect the tracking effect.When the target is far away from the target,the target occupies a small proportion in the field of view of the image;on the contrary,there will be a large proportion,and the edge contours and texture information are clearly distinguishable.In this paper,a vision-based target tracking method for UAV is proposed.In order to reduce the negative effects of interference in the background environment,this paper attempts to study the effective means of long-time tracking and large-scale target detection.The aim is to improve the real-time performance of the system and robustness.The main work of the thesis is as follows:1、Firstly,it introduces the research background of current domestic and foreign tracking algorithms,and makes a basic explanation for the current main detection and tracking systems.At the same time,on the target detection and tracking algorithm level,it analyzes the advantages and disadvantages and development direction between them,and explores The current research problems,find the place to be broken,and thus lay the direction for the research work of this paper.2、A particle filter tracking algorithm based on background weighted local sensitive histogram(LSH)combined with edge gradient histogram(HOG)are proposed.This method can enhance the description of the tracked target features,adapt to target scale changes,target occlusion,etc.And illuminance invariant features.It can well realize the detection and tracking of small targets,The problem of occlusion,scale and illumination changes in the moving target tracking process in outdoor complex environments is effectively solved.3、A deflection strategy is designed,which is based on the position relation between UAV,camera and target,and uses flight state data to carry out feedback control.Which obtains the image location information of UAV by airborne sensor,establishes the relationship between UAV posture information and ground target size,and makes the tracking target always in the image center of the drone cloud camera.At the same,in order to verify the performance of the tracking algorithm.The thesis designs and implements the simulation platform of the UAV ground target tracking system and carries out the indoor simulation experiment,and carries out the outdoor moving target(the target is the person in the paper)tracking experiment.The results show that the proposed algorithm can not only successfully follow the human motion,but also keep the camera of the drone always in the direction of the target,and ensure that the moving target is always at the center of the image.