Object-Oriented Interpretation and Classification with Airborne LiDAR 3D Data

Abstract

Three dimensional (3D) information acquisitions is one of the advantages of airborne Light Detection and Range (LiDAR) as interpretation data sources and the relative classification and feature extraction algorithms based on the point cloud height are developed. In this paper, in the study site of Lac-Saint-Charles, Quebec, Canada, the 3D accuracy of the airborne LiDAR products, including digital elevation model (DEM), digital surface model (DSM) and intensity image, are evaluated with differential GPS (dGPS) data. An object-oriented interpretation and classification work flow with these products is proposed and carried out: terrain group and up-terrain group are acquired based on the digital height model (DHM) and accuracy assessment result with dGPS data, five different classes including water body, impervious areas, bare soil, vegetation and building are separated based on a multi-window homogeneity Maximum Likelihood Supervised (MLS) classification. Based on accuracy statistics with 965 independent checked data, the final classification result and experimental accuracy analysis demonstrated an efficient attempt in land cover classification with airborne LiDAR products.