In: Paparoditis N., Pierrot-Deseilligny M.. Mallet C. Tournaire O. (Eds). 1APRS. Vol. XXXVIII. Part ЗА - Saint-Mandé, France. Septeniber 1-3. 2010 
304 
4. CONCLUSIONS AND OUTLOOK 
In this paper, an approach for the extraction of a road network 
in suburban areas was presented. CIR images and a DSM were 
used to first segment an image, then extract road parts and 
connect them to finally form a road network. The results 
presented in this paper show that the approach is suitable for the 
extraction of a road network in a suburban scene. From all 
examined subsets, three quarters of the road network could be 
extracted, and more than 90% of the extracted roads were 
correct. The approach was tested on two different data sets 
(Grangemouth and Vaihingen). Despite the fact that the two 
data sets had quite different sensor characteristics, we used 
identical parameters for our road extraction algorithm, with the 
exception of the NDVI threshold that had to be adapted 
manually. This suggests that the parameter set is quite robust; 
however, a further sensitivity analysis would be desirable. 
Whereas the total completeness was lower in the Vaihingen data 
set (mainly because the examined subsets there contained more 
roads covered by trees), the correctness was consistently good, 
which shows that the approach can be used for images from 
different sensors and different suburban areas. An important 
aspect to be improved is the geometric accuracy. This concerns 
several parts of the algorithm. The extraction of the centre lines 
from the irregularly shaped road parts could be improved by a 
previous orientation-dependent smoothing of the road parts. 
The junctions could be more explicitly modelled and their 
verification could be enhanced by using context objects in a 
similar way to that used for the subgraphs. When the network is 
extracted, the geometric positions of the roads could be 
improved using a snake-based algorithm. The completeness of 
the network could be improved by a search for gaps in the 
network and an evaluation of these gaps, e.g. by examining 
valleys in the DSM starting from road ends. 
ACKNOWLEDGEMENTS 
The Vaihingen data set was provided by the German 
Association for Photogrammetry and Remote Sensing (DGPF): 
http://www.ifp.uni-stuttgart.de/dgpf/DKEP-Allg.html. The 
normalized cuts calculation was adapted from a Matlab program 
from T. Cour, S. Yu and J. Shi: http://www.seas.upenn.edu/ 
~timothee/software_ncut/software.html (accessed May 2010). 
The linear program was calculated with the LP solver lp_solve: 
http://lpsolve.sourceforge.net/5-5/ (accessed May 2010). 
REFERENCES 
Baumgartner, A., Steger, C., Mayer, H., Eckstein, W. and 
Ebner, H., 1999. Automatic road extraction based on multi 
scale, grouping, and context. Photogramm. Eng. Rem. S. 65(7), 
pp.777-785. 
Cramer, M. and Haala, N., 2009. DGPF Project: Evaluation of 
digital photogrammetric aerial based imaging systems - 
overview and results from the pilot centre. In: IAPRSSIS 
XXXVIII( 1-4-7/W5), on CD-ROM. 
Dantzig, G.B., 1963. Linear programming and extensions. 
Princeton University Press, Princeton, New Jersey, USA. 
Doucette. P., Agouris, P., Stefanidis, A. and Musavi, M., 2001. 
Self-organized clustering for road extraction in classified 
imagery. ISPRS J. Photogramm. 55(5-6), pp. 347-358. 
Gerke, M. and Heipke, C, 2008. Image-based quality 
assessment of road databases. Int. J. Geogr. Inf. Sei., 22(8), pp. 
871-894. 
Grote, A., Butenuth, M. and Heipke, C., 2007. Road extraction 
in suburban areas based on normalized cuts. In: IAPRSSIS 
XXXVI(3/W49A) pp. 51-56. 
Grote, A. and Heipke, C., 2008. Road extraction for the update 
of road databases in suburban areas. In: IAPRSSIS 
XXXVII(B3b), pp. 563-568. 
Grote, A., Heipke, C., Rottensteiner, F. and Meyer, H., 2009. 
Road extraction in suburban areas by region-based road 
subgraph extraction and evaluation. In: Urban Remote Sensing 
Joint Event, Shanghai, 20-22 May, on CD. 
Hinz, S. and Baumgartner, A., 2003. Automatic extraction of 
urban road networks from multi-view aerial imagery. ISPRS J. 
Photogramm. 58(1-2), pp. 83-98. 
Hu, X., Tao, C.V. and Hu, Y., 2004. Automatic road extraction 
from dense urban area by integrated processing of high 
resolution imagery and LIDAR data. In: IAPRSSIS 
XXXV(B3), pp. 288-292. 
Hu, J., Razdan, A., Femiani, J.C., Cui, M. and Wonka, P., 2007. 
Road network extraction and intersection detection from aerial 
images by tracking road footprints. IEEE TGARS 45(12), pp. 
4144-4156. 
Mena, J.B. and Malpica, J.A., 2005. An automatic method for 
road extraction in rural and semi-urban areas starting from high 
resolution satellite imagery. Pattern Recogn. Lett. 26(9), pp. 
1201-1220. 
Poullis, C. and You, S., 2010. Delineation and geometric 
modeling of road networks. ISPRS J. Photogramm. 65(2), pp. 
165-181. 
Shi, J. and Malik, J., 2000. Normalized cuts and image 
segmentation. IEEE TP AMI 22(8), pp. 888-905. 
Shackelford, A.K. and Davis, C.H., 2003. Fully automated road 
network extraction from high-resolution satellite multispectral 
imagery. In: Proc. IGARSS, Toulouse, Vol. 1, pp. 461-463. 
Wiedemann, C., Heipke, C., Mayer, H. and Jamet, O., 1998. 
Empirical evaluation of automatically extracted road axes. In: 
K.J. Bowyer and P.J. Phillips (eds.): Empirical Evaluation 
Methods in Computer Vision, IEEE Computer Society Press, 
Los Alamitos, California, pp. 172-187. 
Wiedemann, C. and Ebner, H., 2000. Automatic completion and 
evaluation of road networks. In: IAPRSSIS XXXIII(B3/2) pp. 
979-986. 
Youn, J. and Bethel, J.S., 2004. Adaptive snakes for urban road 
extraction. In: IAPRSSIS XXXV(B3) pp. 465-470. 
Zhang, C., 2004. Towards an operational system for automated 
updating of road databases by integration of imagery and 
geodata. ISPRS J. .Photogramm. 58(3-4), pp. 166-186. 
Zhang, Q. and Couloigner, L, 2006. Automated road network 
extraction from high resolution multi-spectral imagery. In: Proc. 
ASPRS Annual Conf., Reno, Nevada, 10 p., on CD-ROM.