一种投影点背景判断的NLOS卫星检测方法  

A detection method for NLOS satellite based on background judgment of projection points

作　　者：马家能 于先文[1] 马德皓 MA Jianeng;YU Xianwen;MA Dehao(School of Transportation,Southeast University,Nanjing 211189,China)

机构地区：[1]东南大学交通学院,南京211189

出　　处：《导航定位学报》2025年第2期111-117,共7页Journal of Navigation and Positioning

基　　金：国家重点研发计划项目(2022YFB4300300)。

摘　　要：针对视觉辅助检测非视距(NLOS)卫星时所遇到的计算量大、准确度不够的问题,提出一种基于卫星投影点误差椭圆背景判断的非视距卫星快速检测方法:相机朝向上方拍摄图像,将卫星投影至相机图像上,计算卫星在图像上投影点的像素坐标与误差椭圆;通过神经网络对误差椭圆内图像的背景进行判断,若判断结果显示为天空,则推断该卫星信号为直接视距信号,否则为非视距信号。实验结果表明,检测并剔除NLOS信号后,卫星定位精度可提升41.4%;同时,在检测时间和资源占用上优于现有方法,可分别减少87.1%和97.1%。Aiming at the problem of large computation and insufficient accuracy encountered in visually assisted detection of non-line-of-sight(NLOS)satellites,the paper proposed a fast detection method for NLOS satellite based on the background judgment of the error ellipse of the satellite projection point:the camera shot the image facing upwards,the satellite was projected onto the camera image,and the pixel coordinates and error ellipse of the satellite projection point on the image were calculated;then,the background of the error ellipse was judged by the neural network,and the satellite signal was considered to be a line-of-sight signal if it is shown to be sky,otherwise it is a NLOS signal.Experimental results showed that the accuracy of the satellite positioning could be improved by 41.4%after detecting and removing NLOS signals;meanwhile,the method would outperform the existing methods in terms of detection time and resource consumption,which could be reduced by 87.1%and 97.1%,respectively.

关 键 词：卫星定位 非视距(NLOS)信号 卫星投影 误差椭圆 天空识别 

分 类 号：P228.4[天文地球—大地测量学与测量工程]