Photogrammetric refinement of LiDAR-derived building roof contours

Abstract

In this paper, a photogrammetric method is proposed for refining 3D building roof contours extracted from airborne Light Detection And Ranging (LiDAR) data. It is assumed that LiDAR-derived planar faces of roofs are potentially accurate, while LiDAR-derived building roof contours are not well defined. First, polygons representing building roof contours are extracted from a high-resolution aerial image. Second, straight-line segments delimitating each building roof polygon are projected onto the corresponding LiDAR-derived roof planes by using a line-based photogrammetric model. Finally, refined 3D building roof contours are reconstructed by connecting every pair of photogrammetrically projected adjacent straight lines. The obtained results showed that the proposed approach worked properly, meaning that the integration of image data and LiDAR data allows better results to be obtained, when compared to the results generated by using only LiDAR data.