Research on the terrain suitability of the ICCP algorithm

The matching performance of terrain-aided navigation system is directly dominated by terrain features. Therefore, it is not only the requirement of estimating the matching performance but also the basis of planning a trajectory to model the quantitative relationship of the terrain features to the matching algorithms. In this paper, the relationship of terrain features to the iterative closest contour point (ICCP) algorithm is studied by simulations. Firstly, 6 terrain factors are selected from the 17 ones according to the cluster and correlation analysis. Then, the multiple regression method, the logistics analysis method and the discriminant analysis method are applied to derive the three different quantitative formulas between the ICCP matching errors and the terrain factors. The multiple regression method aims to fit the matching errors with the terrain factors explicitly, while the logistics analysis method and the discriminant analysis method try to judge the relationship between groups of matching errors and the corresponding terrain factors. Simulation results show that the prediction accuracy with the multiple regression method coincides with the real basically, but there are very large discrepancy at some points. Simulation results also prove that the formulas derived by the logistics analysis method and the discriminant analysis method can judge all of the 252 samples at the more than 90% correctness.