ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., II-3/W5, 113-119, 2015
http://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/II-3-W5/113/2015/
doi:10.5194/isprsannals-II-3-W5-113-2015
© Author(s) 2015. This work is distributed
under the Creative Commons Attribution 3.0 License.
 
19 Aug 2015
EVALUATING THE POTENTIAL OF MULTISPECTRAL AIRBORNE LIDAR FOR TOPOGRAPHIC MAPPING AND LAND COVER CLASSIFICATION
V. Wichmann1,2, M. Bremer2,3, J. Lindenberger4, M. Rutzinger3,5, C. Georges1,2, and F. Petrini-Monteferri1 1Laserdata GmbH, Technikerstr. 21a, 6020 Innsbruck, Austria
2alpS GmbH, Centre for Climate Change Adaptation, Grabenweg 68, 6020 Innsbruck, Austria
3Institute of Geography, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria
4TopScan GmbH, Düsterbergstr. 5, 48432 Rheine, Germany
5Institute for Interdisciplinary Mountain Research, Austrian Academy of Science, Technikerstr. 21a, 6020 Innsbruck, Austria
Keywords: Multispectral LiDAR, spectral patterns, pattern matching, automatic classification Abstract. Recently multispectral LiDAR became a promising research field for enhanced LiDAR classification workflows and e.g. the assessment of vegetation health. Current analyses on multispectral LiDAR are mainly based on experimental setups, which are often limited transferable to operational tasks. In late 2014 Optech Inc. announced the first commercially available multispectral LiDAR system for airborne topographic mapping. The combined system makes synchronic multispectral LiDAR measurements possible, solving time shift problems of experimental acquisitions. This paper presents an explorative analysis of the first airborne collected data with focus on class specific spectral signatures. Spectral patterns are used for a classification approach, which is evaluated in comparison to a manual reference classification. Typical spectral patterns comparable to optical imagery could be observed for homogeneous and planar surfaces. For rough and volumetric objects such as trees, the spectral signature becomes biased by signal modification due to multi return effects. However, we show that this first flight data set is suitable for conventional geometrical classification and mapping procedures. Additional classes such as sealed and unsealed ground can be separated with high classification accuracies. For vegetation classification the distinction of species and health classes is possible.
Conference paper (PDF, 1239 KB)


Citation: Wichmann, V., Bremer, M., Lindenberger, J., Rutzinger, M., Georges, C., and Petrini-Monteferri, F.: EVALUATING THE POTENTIAL OF MULTISPECTRAL AIRBORNE LIDAR FOR TOPOGRAPHIC MAPPING AND LAND COVER CLASSIFICATION, ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., II-3/W5, 113-119, doi:10.5194/isprsannals-II-3-W5-113-2015, 2015.

BibTeX EndNote Reference Manager XML