火星局部地形高精度配准方法研究

High Precision Registration Method for Local Terrain of Mars

针对HiRISE(high resolution science imaging experiment)DEM(digital elevation model)和MOLA(Mars orbiter laser altimeter)DEM的匹配问题,提出了一种基于点云粗配准与精配准方法结合的地形匹配框架。实验表明,相比于GICP等经典方法,本文算法在“好奇号”着陆区域的地形配准精度提高近20%,且收敛速度更快,算法鲁棒性更强。

For the matching of HiRISE DEM and MOLA DEM,a terrain matching framework based on the combination of point cloud coarse registration and fine registration is proposed in this paper.Firstly,we use the nearest neighbor matching of ISS+FPFH feature space to locate the global coarse location of HiRISE DEM.Secondly,in order to solve the problem of large resolution difference between MOLA DEM and HiRISE DEM,we improve the VGICP method.By weighting the loss and covariance by Euclidean distance between distributions,we construct a smooth weighting method considering specificity,which reduces the sensitivity of voxel radius selection to the results and improves the robustness of the algorithm.Experiments show that compared with the classical methods such as GICP,the terrain registration accuracy of our method on the landing area data of Curiosity is improved by nearly20%,and the convergence speed is faster and the robustness of the algorithm is stronger.