ult 
ng 
om 
rea 
but 
ie} 
the 
map data which did not match well with the 
source images. However, this change 
detection failed when the change was 
relatively small compared to the building 
size. 
An expert system approach permitted control 
of the iterations required for feature 
extraction and the refinement of threshold 
values. Transparent nature of the knowledge 
coded as rules in text format allowed easy 
access and understanding. Processing for 
uncertain regions and merged features was 
well-controlled by the expert system using 
the focusing mechanism. Another advantage 
of expert system approach was an ability to 
efficiently find solutions from the large 
descriptor space. 
The experiments of this study demonstrated 
that two to four times smaller pixel 
resolution was required to achieve machine 
feature extractions comparable to those of 
human interpreters. This relationship 
implies the requirement for small pixels 
for automatic feature extraction. As the 
USGS 1:24,000 scale maps show buildings as 
small ae 12.x 12m (USCS, 1961), "and an 
original image pixel resolution of about 5 
m is necessary for human interpreters to 
extract these buildings, the pixel 
resolution required for automatic feature 
extraction will be on the order of 2.5 to 
1.28 m. 
According to the estimation by Light (1986) 
the optimum pixel resolution for a 
cartographic database for 1:24,000 
topographic maps and digital gray-scale 
images is about 2.0 m. Hence, it will be 
possible to extract most buildings required 
for the maps from the images in such 
cartographic databases using automatic 
feature extraction. However, there are 
other map features with smaller dimensions 
than buildings, e.g. narrow roads and 
creeks. For these features, smaller pixel 
resolution images may have to be resampled 
from the images in the database. 
Consequently, automatic extraction of such 
small features will require larger data 
Storage and longer processing time. 
ACKNOWLEDGEMENTS 
Authors would like to acknowledge the use 
of the SPOT image data employed in this 
article. These data are copyrighted by 
CNES, Toulouse, France. 
REFERENCES 
Bajcsy, R., and M. Tavakoli. 1976. 
"Computer Recognition of Roads from 
Satellite Pictures," IBEF Trans. on 
Systems, Man, and Cybernetics, Vol. 
SMC-6, No.9,; pp. 623-637. 
Buchanan, B.G., and E.H. Shortliffe. 1984. 
Rule-Based Expert Systems, Addison- 
Wesley Publishing Co., Reading, MA. 
Fischler, M.A., J.M. Tennenbaum, and H.C. 
Wolf. 1981 "Detection of Roads and 
Linear Structures in Low-Resolution 
Aerial Imagery Using a Multisource 
Knowledge Integration Technique," 
Computer Graphics and Image Processing, 
Vol. 15, pp. 201-223. 
Hanson, A., and E. Riseman. 1988. "The 
Visions Image Understanding System," 
Advances in Computer Vision : Vol. 1, 
Ed. by C. Brown, Lawrence Erlbaum 
Associates, Hillsdale, NJ. 
Huertas, A., and R. Nevatia. 1988. 
"Detecting Buildings in Aerial Images," 
Computer Vision, Graphics, and Image 
Processing, Vol. 41, pp. 131-152. 
Light, D. L. 4986. "Planning for Optical 
Disk Technology with Digital 
Cartography," Photogrammetric 
Engineering and Remote Sensing, Vol. 
52, PD 551-557 
Lillesand, T.M., and R.W. Kiefer. 1987. 
Remote Sensing and Image 
Interpretation, John Wiley and Sons, 
Inc. New York, NY. 
Matsuyama, T., and V.S-S. Hwang. 1990. 
SIGMA: A Knowledge-Based Aerial Image 
Understanding System, Plenum, NY. 
McKeown, D.M., W.A. Harvey, and J. 
McDermott. 1985. "Rule-Based 
Interpretation of Aerial Imagery," IEEE 
Trans. Pattern Recognition and Machine 
Intelligence, Vol. PAMI-7, pp. 570-585. 
Murakami, H. 1990. Automatic Feature 
Extraction for Map Revision, Ph.D. 
Dissertation, Department of Geography, 
University of Georgia, Athens, Georgia, 
236p. 
Nagao, M., and T. Matsuyama. 1980. A 
Structural Analysis of Complex Aerial 
Photographs, Plenum, NY. 
Nevatia, R., and K.R. Babu. 1980. "Linear 
Feature Extraction and Description," 
Computer Graphics and Image Processing, 
Vol. 13, pp. 257-269. 
Paine, D.P. 1981. Aerial Photography and 
Image Interpretation for Resource 
Management, John Wiley & Sons, New 
York, NY. 
USGS. 1961. "Building and Urban Areas," 
Chapter 3A2 of Topographic Instructions 
of the United States Geological Survey, 
U.S. Geological Survey, Reston, 
Virginia. 
Wang, F., and R. Newkirk. 1988. "A 
Knowledge-Based System for Highway 
Network Extraction," IEEE Trans. on 
Geoscience and Electronics, Vol. GE-14, 
pp. 37-44.