Research On Image Matching Algorithm Of UAV Target Positioning In Night Vision Environment

With the continuous progress of science and technology in recent years,drone technology has developed rapidly,and its target positioning function has played an important role in many fields such as military applications,forestry inspections,and commercial logistics.Target positioning can be achieved by image matching between drone aerial visible light images and visible light satellite imagery images.However,in actual applications,visible light images taken by drones are easily affected by the external environment,and the sharpness of the image is poor in night vision.It affects the accuracy of target localization and can't even locate it.Therefore,this paper studies an image matching algorithm that can accurately locate UAV targets in night vision environment.This article uses drones to capture infrared images and visible light gallery images for matching and positioning.During the positioning process,the gray level difference between the two heterogeneous images of the visible light image and the infrared image is large,and the image matching results are usually not ideal.The characteristics of the edge characteristics of the source image are similar.By extracting the edge image of the heterogeneous image,the effect of gray difference on image matching is reduced,and the Lo G edge detection algorithm is improved.The weighted mean filtering process is added to obtain a more anti-noise capability.Strong,detailed extraction of more accurate edge detection algorithms effectively improves the accuracy of target positioning;due to the large angle difference between the aerial image of the drone and the reference image,the accuracy of traditional image matching is low,and the ORB algorithm is performed.To improve,construct a perspective transformation model,transform the real shot image into a simulated image,eliminate the difference in perspective,and then perform ORB image matching to obtain an algorithm with higher matching accuracy,further improving the accuracy of target positioning;for image matching The algorithm calculation process is more complicated and cannot meet the drone goals Real-time position,the matching process into coarse matching and fine matching two stages,a rough matching result generating homography matrix is regenerated into the simulated perspective image precise matching stage,increasing the speed of the positioning.The experimental results show that the UAV target positioning image matching algorithm under the night vision environment studied in this paper has higher positioning accuracy and better real-time performance.