플래시 라이다를 사용한 템플릿 매칭 기반의 지형참조항법

Abstract

This thesis compares and analyzes a performance of template matching based terrain referenced navigation (TMTRN) using correlation functions according to different error types and correlation functions. Conventional batch processing TRN generally utilizes the radar altimeter and adopts mean square difference (MSD), mean absolute difference (MAD), and normalized cross correlation (NCC) for matching a batch profile with terrain database. If a flash LiDAR is utilized instead of the radar, it is possible to build a profile in one-shot. A point cloud of the flash LiDAR can be transformed into 2D profile, unlike a vector profile obtained from batch processing. Therefore, by using the flash LiDAR we can apply new correlation functions such as image Euclidean distance (IMED) and image normalized cross correlation (IMNCC) which have been used in computer vision field. The simulation result shows that IMED is the most robust for different types of errors.