期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2005
卷号：XXXVI-8/W27
出版社：Copernicus Publications
摘要：In this paper we report on development and testing of some improvements to recently introduced object extraction tech- niques suitable for 3D remotely sensed data such as LIDAR or InSAR. Typical urban entities such as buildings, trees, roads, etc. are characterized based on their 3D structure without a need for referencing a given model. Regularisation and segmentation algorithms are applied to the original Digital Surface Model to remove noise and other artifacts; then, best-fitting-plane criteria are applied to the cleaned data in order to partition the scene into a set of planar patches which can constitute the base element for a subsequent, model-driven, recognition and refinement step. Results are shown on a LIDAR data set over the city of Parma in Northern Italy
关键词：Urban remote sensing; LIDAR; building extraction; best plane fitting
