News from the Software Side:
Filtering DTM and Vegetation in RiSCAN PRO
Located in the southern province of Styria, a mountainous area provided the scene for a perfect test site for Airborne Laser Scanning data. Due to fair weather conditions and minor snow coverage an excellent dataset was captured with the high performance RIEGL airborne laser scanner system NP680i mounted on RIEGL’s DA42 Multi Purpose Platform by end of December.
This dataset can now be used for various tests and post processing purposes.
As a first approach the data was used to assess the possibilities in digital terrain modelling and forest applications, where we see a lot of potential for ALS data in the future. "Due to the rising awareness of the importance of a responsible treatment of our natural resources, we register an increased demand for highly accurate survey data in the forest industry", says Jürgen Nussbaum, RIEGL Director International Sales.
According to our clients it is not unusual to combine terrestrial laser scan data with airborne scan data. Therefore the post processing described below was performed entirely in RiSCAN PRO, to demonstrate the excellent capabilities of this software package to process ALS data, too. The entire project aimed at the visualization of the digital terrain model, roads, trails and vegetation. The scandata was imported to the software with suitable coordinate reduction constants applied, to speed up computation time.
The further post processing was split into the following consecutive tasks:
Digital Terrain Model
For the computation of the digital terrain model (DTM) a separation between ground points and non-ground points had to be done first. There are several ways to deal with that issue, depending on the various information content of the dataset. For the actual data a geometric approach had been chosen. The strategy was to find the measurement with the lowest z-coordinate in a certain computation cell. These points defined a rough terrain model, forming a base to calculate a terrain layer of a certain thickness. The volumetric hull of that layer was then used to separate ground from non-ground points, by means of an inside-outside test. The inside points were the new input dataset for a refined DTM filtering. By looping through this iterative workflow it was possible to refine the raster cells step by step, resulting in a very detailed DTM. For the entire project area (approx. 1400x1600m) a 50cm gridded DTM could be calculated without any decimating algorithms applied. Nevertheless, for larger project areas, some tiling, triangle or resolution reduction should be considered. Finally the filtered set of ground points was fed into a triangulation algorithm to create the mesh output. The DTM model may be regarded as reference base for the further post processing.´
Filtering of Vegetation
The next step was the filtering of vegetation. The applied strategy was to filter equidistant vegetation layers according to the underlying terrain surface. Again this can be seen as a purely geometric approach applicable for the vast majority of scan data available nowadays. For this project we restricted the filtering to two vegetation layers, namely a low vegetation layer (shrub layer) containing all points up to 1 m above ground level (AGL) and a high vegetation layer containing all points above 1m AGL. The function "surface comparison" available in RiSCAN PRO provided the right tool to filter the data in the desired way. The results can be seen in the images besides. It is interesting to note, that the low vegetation layer gives a good indication of the amount of deadfall in a certain area. The high vegetation layer was then used to calculate the canopy surface. A filter was applied to find the local maxima in the raster cells. These points were used to calculate the canopy height model. A difference model between canopy surface and DTM gives a good indication of the tree heights.
Finally some polylines have been digitized on the terrain model to visualize the network of roads and forest trails, where the lineweight indicates the priority of the roads.
The geometric approach should be applicable for most available scandata. Needless to say, that the actual datasets provide much more information, than pure 3D geometry. Ground and vegetation filtering from full waveform analysis is a hot topic in current research projects and RIEGL LMS is successfully contributing its variety of airborne scanners to this interesting field of application.