항공 라이다 데이터를 이용한 건물의 윤곽선 추정

Delineation of the accurate building boundaries is crucial to provide reliable spatial products such as digital maps and city models. In LiDAR(Light Detection and Ranging) data, physical boundaries of the building exist somewhere between outer-most points on the roofs and surrounding points on the ground. The positional accuracy of the boundaries is influenced by ground sampling distance, and only a few points could be obtained from the sidewalls because airborne laser systems collect data from near-nadir direction. This paper presents a method to estimate physical building boundaries from the LiDAR data. Firstly, outer-most points on the buildings were identified, then the nearest points on the ground were determined at every point along the outer-most boundaries. Secondly, Midpoints were computed using outer-most points and the nearest points. Finally, refinement processes including model key point detection and regularization were performed since the midpoints represent approximation of the physical boundaries.