Multivariate Gaussian Decomposition for Multispectral Airborne Lidar Data Classification

Abstract

Multispectral airborne LiDAR technology has been available since 2014 by the first commercial multispectral airborne LiDAR sensor, Optech Titan. The sensor acquires LiDAR data at three independent wavelengths (1550, 1064 and 532 nm). This allows for the collection of a diversity of spectral information from different land objects. Recent studies have been devoted to use the spectral information of the LiDAR data along with the elevation information for classification purposes. In this paper, we present an automatic classification method for multispectral airborne LiDAR data based on the multivariate Gaussian decomposition (MVGD). A data subset covering an urban area in Oshawa, Ontario, Canada was used to test the method. The proposed method achieved an overall accuracy of 95.6% for classifying the multispectral LiDAR data into four different classes.