International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
  
with SFS and “non ground” with TerraScan (see Figure 10): 
this is in agree with results obtained for simulated data. 
  
9000 
8000 +—— 
7000 = 
6000 + 
5000 
4000 4—— 
3000 = 
2000 
1000 
BSFS 
Bl TerraScan 
  
  
Ground Points Non-Ground Points 
Figure 10: SFS vs. TerraScan classification of Gorizia points. 
  
  
About the location of the two classifications, while they 
substantially correspond for building and vegetation zones, the 
disparities mainly occur for roads and parking areas. 
  
edi : ^ 2 E : i B 
Figure 11: LIDAR DSM of the city of Gorizia (Italy). 
  
  
    
Figure 11 shows the DSM obtained with such LIDAR data. The 
results of filtering/classification by SFS algorithm are painted 
over with green spots when ground classified, red spots 
otherwise (outliers). They seem to be very truth, so the SFS 
correctness is qualitatively proved, being very hard to exactly 
evaluate it quantitatively (cars positions are unknown/variable). 
6. CONCLUSIONS AND FUTURE PERSPECTIVES 
This paper illustrates a new robust technique for the filtering of 
non-ground measurements from airborne LIDAR data. The 
algorithm represents an efficient method for automatic 
classification of LIDAR data, mainly based on a newly 
developed tool for robust regression analysis and robust 
estimation of location and shape. The main advantage of using 
SAR models and BFS algorithm relies not only on its accuracy 
but also on its statistical robustness. It makes possible to 
efficiently and simultaneously detect either trend surface or 
outlier points by suitably enlarging a data subset. 
Through a significant number of examples, the paper shows 
how the proposed SFS method is a valuable tool for the purpose 
of filtering LIDAR height measures and terrain modeling. Here, 
examples of rough terrain were processed, showing that the 
method can deal with dataset containing many break lines. 
Besides the Gorizia dataset shown in this paper, the method is 
currently being applied to other large urban datasets of 
increasing point density, for the goal of building extraction also. 
In a near future, to improve the efficiency of the SFS algorithm, 
other space interaction models will be tested, as second order 
SAR models and Conditional AutoRegressive models (CAR). 
    
200 
REFERENCES 
Atkinson, A.C., Riani, M., 2000. Robust Diagnostic Regression 
Analysis. Springer, New York. 
Axelsson, P., 2000. DEM Generation from laser scanner data 
using TIN adaptive models. International Archives of 
Photogrammetry & Remote Sensing , vol. XXXIII, part B3, pp. 
85-92. 
Brovelli, M.A., et aL, 2001. Modelli matematici del terreno per 
mezzo di interpolatori a spline. Bollettino SIFET, supplemento 
speciale al n. 2, pp. 55-80. 
Cerioli, A., Riani, M., 2003. Robust methods for the analysis of 
spatially autocorrelated data. Journal of the Italian Statistical 
Society, 11, pp. 334-358. 
Griffith, D.A., Layne L.J, 1999. A casebook for spatial 
statistical data analysis. Oxford University Press, New York. 
Haining, R., 1990. Spatial data analysis in the social and 
environmental sciences. Cambridge University Press, Cambridge. 
Jacobsen, K., 2001. New Developments in Digital Elevation 
Modelling. Geoinformatics, June, pp. 18-21. 
Kraus, K., 1997. Eine neue Methode zur Interpolation und 
Filterung von Daten mit schiefer Fehlerverteilung. Österreichische 
Zeitschrift für Vermessung & Geoinformation 1, pp. 25-30. 
Kraus, K., Pfeifer, N., 1997. A new method for surface 
reconstruction from laser scanner data. International Archives 
of Photogrammetry & Remote Sensing, vol. XXXII, part 3- 
2W3, pp. 80-86. 
Pace, R.K., Barry, R., Sirmans, C.F., 1998. Quick computation 
of spatial autoregressive estimators. Journal of Real Estate 
Finance and Economics, pp. 1-12. 
Rottensteiner, F., et al, 2002. LIDAR Activities at the 
Viennese Institute of Photogrammetry and Remote Sensing. 
Proceedings of the 3rd Int. LIDAR Workshop, Columbus. 
Soininen, A., 2003. TerraScan. User Guide. 
Voelz, H., Mathematische Grauwertmorphologie zur Filterung 
Digitaler Oberfláchenmodelle, Diplomarbeit, Universität 
Hannover, http://www.ipi.unihannover.de/html/lehre/ 
diplomarbeiten/2001/2001.htm 
Vosselman, G., 2000. Slope based filtering of laser altimetry 
data. International Archives of Photogrammetry & Remote 
Sensing, vol. XXXIII, part B4, pp. 958-964. 
Whittle, P., 1954. On stationary process in the plane. 
Biometrika, 41, pp. 434-449. 
Zhangh, K., et al, 2003. A Progressive Morphological Filter 
for Removing Non ground Measurements From Airborne 
LIDAR Data. IEEE Transactions on Geoscience and Remote 
Sensing, 41, 4, pp. 872-882. 
ACKNOWLEDGEMENTS 
This work was carried out within the research activities 
supported by the INTERREG IIIA Italy-Slovenia 2003-2006 
project *Cadastral map updating and regional technical map 
integration for the Geographical Information Systems of the 
regional agencies by testing advanced and innovative survey 
techniques". 
    
   
  
  
   
   
   
   
    
   
   
   
   
  
   
   
  
   
    
  
   
  
  
  
  
  
   
   
   
   
    
   
  
   
   
  
   
  
   
    
     
  
  
  
  
    
   
   
  
  
  
  
   
  
  
  
  
  
  
  
* Insi 
KEY V 
ABSTI 
Airborr 
time of 
scanner 
measur 
echoes. 
measur 
is furth 
the typ 
using d 
wavefo 
pre-clas 
and the 
Airborr 
has init 
laser sc 
repetiti 
surface. 
where 1 
the tray 
Earth’s 
related 
topogra 
One ad 
photogt 
as a soi 
laser si 
clouds 
pulses : 
path th 
informe 
provide 
forest c 
constitu 
Since | 
amplifi 
possible 
spectrul 
  
= Con