The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008 
Figure 18 shows rooftop detection results of the entire area. 
There is a building located near the border line of the epipolar 
image that the system can not detect correctly due to missing 
line segments in low level extraction step. The remaining 
buildings are accurately extracted. From the detected 3D 
rooftop, we generate a 3D rendering view, as shown in Figure 
19. The 3D view indicates that the detected 3D rooftop 
accurately reflects ground truth 3D model. 
6. CONCLUSIONS 
A new approach to detect and reconstruct buildings using 
perceptual organization from two aerial images has been 
suggested. Low level feature extraction is performed in the 
epipolar images, which makes it possible to reduce the search 
effort in matching process. The proposed suspected building 
regions can be utilized to remove the unnecessary line segments, 
prior to the generation of rooftop hypotheses. This region can 
reduce computational complexity and false hypotheses. Using 
undirected feature graph, the selection of rooftop hypotheses 
becomes a simple graph searching for close cycles. 
Experimental result shows that the proposed method can be 
very effectively utilized for the rectilinear structures of urban 
area. 
REFERENCES 
Banks, J., Bennamoun, M., Corke, P., 1997. Non-parametric 
technique for fast and robust stereo matching. Proc. of IEEE 
TENCON, pp. 365-368. 
Boldt, M., Weiss, R., Riseman, E., 1989. Token-based 
extraction of straight lines. IEEE Transaction on Systems Man 
Cybernetics, 19(6), pp. 1581-1594. 
Collins, R., Jaynes, C., Cheng, Y.Q., Wang, X., Stolle, F., 
Riseman, E., Hanson, A., 1998. The Ascender system: 
automated site modeling from multiple aerial images. Computer 
Vision and Image Understanding, 72(2), pp. 143-162. 
Fischer, A., Kolbe, T., Lang, F., Cremers, A. B., Forstner, W., 
Plumer, L., Steinhage, V., 1998. Extracting buildings from 
aerial images using hierarchical aggregation in 2D and 3D. 
Computer Vision and Image Understanding, 72(2), pp. 185-203. 
Huertas, A., Nevatia, R., 1988. Detecting buildings in aerial 
images. Computer Vision, Graphics and Image Processing, 
41(2), pp. 131-152. 
Lin, C. A., Nevatia, R., 1998. Building detection and 
description from a single intensity image. Computer Vision and 
Image Understanding, 72(2), pp. 101-121. 
ACKNOWLEDGEMENTS 
This work was supported by the Korea Science and Engineering 
Foundation (KOSEF) grant funded by the Korean government 
(MOST)(Grant No.: R01-2007-000-20330-0). 
Jaynes, C., Stolle, F., Collins, R., 1994. Task driven perceptual 
organization for extraction of rooftop polygons. IEEE 
Workshop on Application of Computer Vision, pp. 152-159. 
Noronha, S., Nevatia, R., 2001. Detection and modeling of 
buildings from multiple aerial images. IEEE Transaction on 
Pattern Analysis and Machine Intelligence, 23(5), pp. 501-518. 
Mohan, R., Nevatia, R., 1989. Using perceptual organization to 
extract 3D structure. IEEE Trans. Pattern Analysis and Machine 
Intelligence, 11(11), pp. 1121-1139. 
Zabih, R., Woodfill, J., 1998. Non-Parametric Local 
Transforms for Computing Visual Correspondence. Proc. Third 
European Conf Computer Vision, pp. 5-28.