Automated extraction of buildings from aerial LiDAR point cloud and digital imaging datasets for 3D cadastre - preliminary results

Abstract

The demand for accurate 3D mapping of buildings has increased due to the spatial detail required by engineers, architects and city planners. An accurate information about location and dimension of building features is important for cadastral, city modeling, infrastructure mapping and safety analysis in an urban environment. LiDAR technology provides rapid, continuous and cost effective capability to acquire 3D geospatial information. In this paper, we present an automated approach for extracting building features from integrated aerial LiDAR point cloud and digital imaging datasets. Our approach is based on the assumption that the LiDAR data can be used to distinguish between high and low rise objects while the multispectral dataset can be used to filter out vegetation from the data. We make a use of LiDAR elevation and multiple echo attributes to extract building objects. Morphological operations are applied to the extracted building objects in order to complete their shapes and remove noise. We tested our automated buildings extraction approach on aerial LiDAR point cloud and digital imaging datasets of Istanbul city. The successful extraction of building objects validates our automated approach.