Impact of terrain factors on the matching performance of terrain-aided navigation

Aiming at describing terrain variation in an easily understandable way, a new method is proposed to describe the terrain elevations quantitatively by factor analysis for terrain-aided navigation. After as many terrain factors as possible are collected describing the terrain elevations from various aspects, they are simplified by correlation to remove the redundant factors. The simplified terrain factors are processed to derive the extracted factors by principal component analysis. With such factor analysis, three extracted factors dubbed “variation,” “similarity,” and “information” are derived in the conceptual level. A classifier is derived by logistic regression using the extracted factors. The percentages of the correct classification are very high both for the suitable and the unsuitable regions of matching in two maps, which proves the effectiveness of the extracted factors for the terrain elevations.