A 
— ———— — nÁ— — P à n 
Y Y 
/ } 
The Separating Hyperplanes method is quite new in the classification of 
MSS data, but several authors of Pattern Recognition textbooks (especially 
Meisel 1972) have discussed the algorithm quite adequately. This author is im- 
pressed by its performance, but appreciates the fact that it, like the Maximum 
Likelihood method, should not be used as a "black box". Investigations should 
therefore continue in the search for the point where the Maximum Likelihood 
should be turned off and the Separating Hyperplanes turned on, or vise-versa, 
in order to obtain the best classification results. 
6. REFERENCES 
Donker, N.H.W. and Mulder, N.J. 1976: Analysis of MSS digital Imagery with the 
aid of Principal Component Transform. Paper presented to the XIII 
Congress of ISP, Helsinki. 
Ekenobi, S.L. 1981: Untersuchungen zur digitalen Landnutzungsklassifizierung 
mit Hilfe von multispektralen Satellitenbildern. Wiss. Arbeiten der 
Fachrichtung Vermessungswesen der Universität Hannover, Nr. 108. Ph.D. 
Dissertation. 
Ekenobi, S.L. 1982: Effect of Differences in Categories Dispersion Patterns on 
Digital Image Classification Results. Paper presented to the Interna- 
tional Geoscience and Remote sensing Symposium, Munich, West-Germany, 
June 1-4, 1982. 
Meisel, W.S. 1972: Computer-Oriented Approaches to Pattern Recognition, Acade- 
mic Press, New York. 
Morrison, D.F. 1976: Multivariate Statistical Methods, 2nd Edition, McGraw-Hill. 
Mulder, N.J. and Hempenius, S.A. 1974: Data Compression and data Reduction 
Techniques for the Visual Interpretation of Multispectral Images, ITC 
Journal 1974-3. 
Swain, P.H. and Davis, S.M. (eds.) 1978: Remote Sensing: The Quantitative Ap- 
proach. McGraw-Hill International Book Company. 
70 
NE uu ub LA m 
In tl 
been 
in vi 
20 G 
turn: 
the 
ance: 
In ri 
iden 
ry (i 
corr 
1981 
Tida 
fors 
and 
corr 
(SNR 
So p 
meth 
On t 
lati 
func 
Gopf 
Int 
tion 
MOBI 
The 
allo 
proc 
b) t 
5 (3 O