tem is 
ion. 
ie line 
it data 
e line 
  
Fig. 10 Representation of the reconstructed toy block of Fig.3 
5 Discussion 
A concept of a vision system with the aim to interpret images 
of polyhedral objects has been presented. It could be shown 
that an interpretation of single images is feasible. 
For solving the interpretation problem, assumptions are made 
about planar faces of the object and the incidence relation 
between the 2D features and hypothesis about spatial 
relationships of the object, being derived either by grouping 
procedures or by the operator. 
The result of the vision system is the 3D reconstruction of a 
polyhedral object from one single image. The input image may 
contain real and artificial objects and graphical sketches as 
well. 
An automatic derivation of buildings or groups of buildings 
may represent an important contribution for the aquisition and 
actualization of information-systems. Applications are city- 
planning in simulated parts of a town and the analysis of the 
climate in cities (BECKROGE/JUNIUS 1991). 
An interpretation of single images of buildings in real-time 
may be implemented in systems for automated navigation of 
cars or planes. The results of the vision system presented here 
are also suited as models for the feature based object location 
in aerial images (SCHICKLER 1992). 
For further applications the system can be transferred to all 
kinds of objects being approximately a polyhedra, e.g. when 
grasping of machine parts in robotik. 
Especially the interpretation of scanned line drawings (inverse 
computer graphics) can be regarded as a part of a man-machine 
interface. In all cases the result of the interpretation of single 
images may used as input for procedures of 3D-CAD graphics. 
Future work will concern the extension to non planar surfaces 
and the fusion of single 3D models. 
6 Literature 
BARNARD, St. (1982): Interpreting Perspective Images. 
Technical Note 271, SRI Project 1009, Menlo Park, 1982 
BECKROGE, W. and JUNIUS, H. (1991): Klimatologische 
Begutachtung bestimmter stadtplanerischer Untersuchungs- 
bereiche im Dortmunder Südwesten unter Einsatz eines Geo- 
Informationssystems. VR 53/3+4, Juni 1991 
BRILLAUT-O'MAHONY, B. (1991): New Method for 
Vanishing Point Detection. CVGIP:IU, Vol. 54, No. 2, 1991, 
pp. 289-300 
CLOWES, M. B. (1971): On Seeing Things. Artificial 
Intelligence 2, 79-116, 1971 
HARALICK, R. M. (1989): Monocular Vision Using Inverse 
Perspective Projection Geometry. Analytic Relations, Conf. on 
Computer Vision and Pattern Recognition, San Diego, 1989, 
pp. 370ff 
HERMAN, M. and KANADE, T. (1986): The 3-D Mosaic 
Scene Understanding System. From Pixels to Predicates, Ablex 
Publishing Corporation, New Jersey, 1986 
HUFFMANN, D. A. (1971): Impossible Objects as Nonsense 
Sentences. Machine Intelligence, Vol.6, ed. by B. Meltzer, D. 
“Michie (Edinburgh University Press, Edinburgh, UK. 1971), 
pp.295-323 
KANATANI, K. (1990): Group-Theoretical Methods in Image 
Understanding, Springer, Berlin 1990 
MAGEE, M. and AGGARVAL, J. (1984): Determining 
Vanishing Points from Perspective Images. CVGIP, Vol. 26, 
1984, pp. 256-267 
MOHAN, R. and NEVATIA, R. (1987): Perceptual Grouping 
for the Detection and Description of Structures in Aerial 
Images. IRIS, Univ. of South Calif., Rep. 225, Los Angeles, 
1987. 
MULGAONKAR, P. G. and SHAPIRO, L. G. (1985): 
Hypothesis-Based Geometric Reasoning about Perspective 
Images. Proc. 3rd Workshop on Computer Vision, Bellaire, 
Michigan, 1985, pp.11-18. 
NEVATIA, R. and ULUPINAR, F. (1991): Constraints for 
Interpretation of Line Drawings under Perspective Projection. 
CVGIP:1U, Vol.53, No.1, 1991, pp.88-96 
NIEMANN, H. (1981): Pattern Analysis, Springer, Berlin 
1981 
PAN, H. P. (1990): A Spatial Structure Theory in Machine 
Vision and its Application to Structural and Textural Analysis 
of Remotely Sensed Images, PhD Thesis, ITC, 1990 
SCHICKLER, W. (1992): Feature Matching for Outer 
Orientation of Single Images using 3D Wireframe 
Controlpoints. ISPRS Congress, COM. III, Washington D.C., 
1992 
SUGIHARA, K. (1986): Machine Interpretation of Line 
Drawings. The MIT Press, Cambridge, MA, 1986. 
STRAFORINI, M., C. COELHO, M. CAMPANI and V. 
TORRE (1990): On the Understanding of Indoor Scenes. 
Intern. Workshop on Robust Computer Vision, Seattle, WA, 
1990 
VOSSELMANN, G. (1991): Relational Matching. Promotion 
Universität Bonn, 1991 
521