143 
4 Summary 
We have described an information fusion approach to 
cartographic feature extraction. We have discussed two 
problems in aerial image interpretation, the fusion of 
monocular scene analysis data and the improvement of three- 
dimensional scene interpretations. We have briefly shown 
the application of information fusion techniques to these 
problems, as well as results for typical suburban aerial 
imagery. We have also discussed potential avenues for 
further research utilizing the information fusion approaches 
described here. 
Cartographic feature extraction from aerial imagery is a 
difficult problem, which requires analysis techniques that 
utilize image domain cues as well as a priori or contextual 
information. The analyses produced by such techniques can 
be complementary; they can also be redundant, or inaccurate. 
Information fusion methods provide a means for the 
integration of such data into a more accurate and 
comprehensive interpretation of the imagery. 
Figure 5: Suburban BABE results 
Figure 7: Suburban SHAVE results 
5 Bibliography 
1. Fua, P., Hanson, A. J., “Resegmentation Using 
Generic Shape: Locating General Cultural Objects”, 
Tech, report, Artificial Intelligence Center, SRI 
International, May 1986. 
2. Mohan, R., Nevada, R., “Using Perceptual 
Organization to Extract 3-D Structures”, IEEE 
Transactions of Pattern Analysis and Machine 
Intelligence, Vol. 11, No. 11, November 1989, pp. 
1121-1139. 
3. Huertas, A. and Nevatia, R., “Detecting Buildings in 
Aerial Images”, Computer Vision, Graphics, and 
Image Processing, Vol. 41, April 1988, pp. 131-152. 
4. Nicolin, B., and Gabler, R., “A Knowledge-Based 
System for the Analysis of Aerial Images”, IEEE 
Transactions on Geoscience and Remote 
Sensing, Vol. GE-25, No. 3, May 1987, pp. 317-329. 
Figure 6: Suburban SHADE results 
Figure 8: Suburban fusion results