ABSTRACT
This chapter presents a unique framework of feature grouping for modeling three dimensional (3D) buildings by employing a Binary Space Partitioning Tree. It discusses the background knowledge pertaining to building reconstruction, a general hierarchy of building reconstruction process, and different types of building object representations. A wider range of building shapes may be difficult to model by relying only on the rectangular partitioning result. The chapter examines a new technique to recover a full delineation of 3D building shape from incomplete geometric features extracted from airborne Light Detection and Ranging (LiDAR) data. The unique characteristics of LiDAR surveying provide great potential through complementarity with optical sensors, for automating the sophisticated tasks involved in 3D cityscape modeling. F. Rottensteiner et al. employed the Dempster–Shafer theory as a data fusion framework for building detection from airborne orthophotos with multispectral bands and LiDAR data. The model-driven approaches fit prespecified models to LiDAR data, thereby determining geometric parameters and topological relations across modeling cues.
