Assessing the Usability of LiDAR Processing Methods on UAV Data

Abstract

By combining stereo images in order to generate accurate point-clouds of the Earth's surface, the UAV (Unmanned Aerial Vehicle) has recently become a cheap alternative to the well-established LiDAR systems. In this paper, we verify to which extent the methods that were developed to process aerial LiDAR data can be used when presented by the limitations of the UAV data (e.g. absence of radiometric information and ground points within vegetated areas). For this purpose, DTM was generated and points were classified for point-clouds that covered identical urban, rural and forested areas but were obtained using two different acquisition methods (i.e. aerial LiDAR and stereo photography from a UAV). The results have shown that the majority of features, utilized by the LiDAR processing methods, are still valid when UAV data is processed. However, the absence of ground points within large clusters of vegetated areas can cause errors during DTM generation and subsequent point classifications.