Discretization of object-based lidar features for land cover classification

摘要

Airborne light detection and ranging (LiDAR) technology is an active remote sensing technique that produces a true orthophoto at a single wavelength. LiDAR is not as dependent on the weather as optical sensors, and LiDAR features have been widely applied for characterizing land cover classes of interest. Instead of using point-based features for classification, object-based LiDAR features were employed in this study. In addition, instead of directly mining a data set with large numerical attributes, a preprocessing step, known as data discretization, was applied before data mining. The results show that although original data are simplified, mining results are concise, more meaningful and easy-to-interpret. In this way, the accuracy of land cover classification can be further improved.