Building Inventory Information Extraction From Remote Sensing Data and Statistical Models

Abstract eng:
In this paper, algorithms for extracting building attribute information from remotely sensed data are presented. In particular, a methodology for rapidly extracting spatial and structural information from a single highresolution satellite image, using rational polynomial coefficients (RPCs) as a camera replacement model is introduced. Geometric information defining satellite’s sensor orientation is used in conjunction with the RPC projection model to generate an accurate digital elevation model (DEM). Additionally, a methodology for inferring engineering attributes of the built-environment, i.e. structural type and occupancy type of buildings, from 3-D building models is formulated. A dataset collected for Southern California, USA, is used to train multinomial logistic regression models and establish inference rules in order to predict the regional engineering parameters of the buildings. Classification error and prediction power of these models are then presented in the paper and an example of the marginal probability distribution computed for a sample building is shown.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket