Research On UAV Positioning And Navigation Technology Based On Image Registration

UAV remote sensing has a very wide range of applications.However,in some electromagnetic shielding environments,GPS signals cannot be relied on to obtain accurate positioning information,and the remote sensing pictures taken cannot be accurately calibrated.This paper studies the visual positioning method based on image registration and three-point positioning under this condition,and the research on this method is carried out in this paper.Among them,the feature-based image registration technology has advantages of fast speed and robustness.In this paper,the distance calibration of the camera is used to solve the relationship between the shooting distance and the image pixel distance and the real distance of the image,and the three-point positioning method is used to reduce the error of the solved shooting distance,and multiple levels are used for the shooting area.In the way of segmentation,the UAV is directed to approach and shoot through path planning to obtain higher-precision regional pictures,and then the local image is registered with the original image,and the UAV is further calculated through the calculation of the space-transformed warp image In order to further update the positioning information of the drone,and then perform fusion processing on the registered pictures,iteratively proceed to obtain the required high-precision pictures finally.The main research contents in this paper are as follows:1)In the case of unknown positioning information,the distance calibration and three-point positioning methods are proposed to achieve the acquisition of UAV position information.2)In the image preprocessing stage,the weighted average method and median filter processing are used.Weighted average method for grayscale processing of the image is more suitable for the grayscale image of the human eye,and reduces the image processing time;the image noise processed by the median filter method is smoother and the noise elimination effect is better.3)In the image registration stage,for the registration method based on the feature point description,the Speeded-Up Robust Features(SURF)descriptor is used.The feature points extracted by this method can accurately represent the visual information in the image,and the matching accuracy of the algorithm is also relatively good.4)In the stage of image fusion,the weighted average fusion algorithm is studied to eliminate the splicing seam and brightness difference in the process of Mosaic,and finally achieve the acquisition of high-precision remote sensing images in a large area.The experimental results show that the method adopted in this paper achieves the acquisition of high-precision wide-field-of-view images while locating the UAV.