Classification of multispectral lidar point clouds

Airborne Light Detection And Ranging (LiDAR) data are widely used for high-resolution land cover mapping. The LiDAR data are typically used as complementary information to passive multispectral or hyperspectral imagery to obtain higher land cover classification accuracy. In this paper, we examine the capabilities of a recently developed multispectral airborne laser scanner, manufactured by Teledyne Optech, for classification of multispectral point clouds into typical land cover classes. This scanner, Titan MW (multi-wavelength), collects point clouds using three different wavelength lasers simultaneously, hence opening the door to new possibilities in land cover classification using only LiDAR data. We show that the recorded intensities of returned laser pulses together with structural characteristics of the features on the Earth surface calculated from the 3D positions of returns are sufficient enough to classify the point cloud into 10 distinct land cover classes. We achieved an overall accuracy of 95.9% with a kappa coefficient of 0.95 using a Support Vector Machine (SVM) classifier to classify single-return points and an overall accuracy of 89.2% and kappa coefficient of 0.82 using a rule-based classifier on multi-return points.