into five 
esent the 
arley 
orn 
ar Beets 
ar Beet> 
tes Trees 
ory and 
ea) 
ing is 
age of 
t, row 
flight 
ve level 
Davy 
t inter- 
me ters 
air 
the 
result 
ne in- 
je ir 
crop 
us 
from 
gained 
ained in 
7 A à 
EEE L e 
the local texture of a SAR image. Although the texture features employed were 
developed in a manner which is rather insensitive to sensor calibration and 
the direction of flight, phenomenological differences in the appearance of a 
specific crop type combined with variations in crop geometry with look direction 
at airborne viewing angles, left me short of the intended goal. Nevertheless, I 
was able to discriminate between crops with field sizes as narrow as 10 meters 
using digital measures of texture in an imagery from a SAR whose ground re- 
solution was 3 meters. In future efforts I plan to work with X,, data, to intro- 
duce additional texture algorithms as well as different texture features and to 
explore techniques for reducing the computational resources required. As LANDSAT 
data processing has benefitted from the use of multitemporal data, I also re- 
commend that multitemporal SAR data be obtained and analyzed for crop discri- 
mination. 
ACKNOWLEDGMENTS 
The author would Tike to express his gratitude to the SWISSAR Investigation 
Team who planned the flight experiment and collected the ground truth infor- 
mation during the day of the overflight. Thanks are also due to a number of 
people at the Environmental Research Institute at Michigan in Ann Arbor, USA: 
Dr.Quentin Holmes and Dr.William Malila for reviewing, Lester Witter and Roger 
Golliver for writing the special software. The images for this report have been 
prepared with an Optronics Film Recorder P-1500 by the Earth Resources Data 
Center at ERIM. 
REFERENCES 
1. Aushermann D.A. et al 1975, Radar Data Processing and Exploitation Facility, 
Proceedings IEEE International Radar Conf., Arlington, Virginia, 
21-23 April 1975, IEEE Publication 75, CHO 938-1 AES, 493-498. 
2. ERIM Report 312300-1-F 1981, ESA/JRC SAR 580 Campaign-Preliminary Imaging 
Data, by J. Marks et al. 
3. Journal of the Optical Society of America, Vol. 66 (1976). 
4. Sternberg S.R. 1979, Automatic Image Processor U.S. Patent 4,167,728, 
Sept. 11, 1979. 
5. Lougheed R.M. et al 1980, The Cytocumputer: A Practical Pipeline Image Pro- 
cessor, 7th Annual Internat.Symp.on Computer Architecture, La Baule, France, 
May 6-8, 1980, IEEE Publication 80CH1494-4C, 271-277. 
6. Sternberg S.R. 1980, Language and Architecture for Parallel Image Processing, 
Proceedings Conference on Pattern Recognition in Practice, Amsterdam, 
May 21-23, 1980. 
7. Preston K. et al 1979, Basics of Cellular Logic with Some Applications in Me- 
dical Image Processing, Proceedings IEEE, Vol.67, No.5, May 1979. 
8. Nüesch D.R. 1982, Augmentation of LANDSAT MSS data by SEASAT-SAR for Agri- 
cultural Applications, Reprint of NASA Rep.(IT-E2-04233) in the Remote 
sensing Series, Vol.7, Dept.of Geogr.,University of Zurich, Switzerland. 
9. Haralick R.M. et al 1973, Textural Features for Image Classification, IEEE 
Trans.Systems, Man & Cybernetics, Vol.SMC-3, Nov. 1973, 610-621. 
729 
ere