基于多帧图像同名点的无人机对地定位新方法

基于单张图像的传统无人机对地定位方法的精度较低，不能满足精确打击作战需求，提出一种新的无人机对地定位方法。该方法通过线性化共线条件方程建立摄像机姿态角和焦距迭代计算模型，根据非共线位置拍摄的多帧图像及其至少3个可识别同名像点坐标迭代计算摄像机姿态角和焦距的精确值，然后利用多摄站前方交会法求取地面目标的三维坐标。该方法不依赖于数字高程模型(DEM)及像片内外方位元素的测量值，消除了传统无人机对地定位方法3个定位误差源中的两个，因此具有较高的定位精度。仿真与真实图像实验计算表明，摄站奉身坐标误差的大小是决定对地定位精度的主要因素，当摄站坐标精度为5m时，对地定位误差约为10 m，远远小于传统方法的定位误差。

A New UAV Geo-location Approach Based on Common Points on Multiple Images

For the accuracy of traditional UAV geo-location approach based on single image is too low to meet the needs of precise strike, new UAV geo-location method was presented. The mathematical models were constructed by linearization of the collinearity equations to iteratively compute the pose angles and focal length of the camera. The camera pose angles and the focal lengths were precisely reckoned by the three images of the target and at least three identifiable common point coordinates a- mong the images. /I hree-dimension coordinates of the ground target were calculated using forward in?tersection. The new method, which is independent of DEM and the measured values of camera pose angles, eliminates the two of the three primary error sources in the traditional UAV target location method and has better location precision. The. results of simulation and real image experiment show that the accuracy of the UAV position is main effective factor on that of the estimated target location, when the former is 5 m, the latter is about 10 m which is much less than that of the traditional.