低空UAV激光点云和序列影像的自动配准方法

提出了一种低空无人机(unmanned aerial vehicle,UAV)序列影像与激光点云自动配准的方法。首先分别基于多标记点过程与局部显著区域检测对激光点云和序列影像的建筑物顶部轮廓进行提取,并依据反投影临近性匹配提取的顶面特征。然后利用匹配的建筑物角点对,线性解算序列影像外方位元素,再使用建筑物边线对的共面条件进行条件平差获得优化解。最后,为消除错误提取与匹配特征对整体配准结果的影响,使用多视立体密集匹配点集与激光点集进行带相对运动阈值约束的ICP(迭代最临近点)计算,整体优化序列影像外方位元素解。试验结果表明本文方法能实现低空序列影像与激光点云像素级精度的自动配准,联合制作DOM精度满足现行无人机产品1∶500比例尺标准。

Automatic Registration of Low Altitude UAV Sequent Images and Laser Point Clouds

It is proposed that a novel registration method for automatic co-registration of unmanned aerial vehicle (UAV)images sequence and laser point clouds.Fi rstly,contours of bui lding roofs are extracted from the images sequence and laser point clouds using marked point process and local salient region detection,respectively.The contours from each data are matched via back-project proximity.Secondly, the exterior orientations of the images are recovered usingal inear solver based on the contours corner pai rs fol lowed by a co-planar optimization which is impl icated by the matched lines form contours pai rs. Final ly,the exterior orientation parameters of images are further optimized by matching 3D points generated from images sequence and laser point clouds using an iterative near the point (ICP)algorithm with relative movement threshold constraint.Experiments are undertaken to check the val idity and effec-tiveness of the proposed method.The results show that the proposed method achieves high-precision co-registration of low-altitude UAV image sequence and laser points cloud robustly.The accuracy of the co-produced DOMs meets 1∶500 scale standards.