CLASSIFICATION OF HEALTHY AND DISEASED SUGARBEETS
WITH REGARD TO OBSERVATIONANGLE EFFECTS
by
PAUL REICHERT
University of Freiburg
Werderring 6
D 7800 FREIBURG
RFA
Airborne MSS-data of a sugarbeet-cultivated area near Frank-
furt, F.R.G. have been analysed to determine the potential of
mutlispectral remote sensing and classification for the detection
various diseases in sugar beets. The analysis has been done in
the framework of & national research programme and an EEC-Study.
Spectral signatures of healthy and diseased sugar beets were de-
termined in situ with a mobile ground station and were calculated
from the row MSS-data after absolute calibration of the scanner.
From both platforms, healthy and diseased sugarbeets have shown
strong differences in their spectral behaviour, which allowed
for classifícation. However, & strong influence of the observation-
zenith angle anü the azimuthangle difference (sun- - observation-
azimuthangle) on the Spectral signatures has been observed. In
order to enhance the classification results, a model of the
snbricat reflection behaviour of sugar beets has been developed
which allowed for the selection of the optimum azimuthangle
difference (showing minimum observation angle distortions)
and the determination of a polynomial function to elminate
observation angle effects. The implementation of this poly-
nomial function resulted in a coniderable gain in classification
accuracy.
The paper should be presented in WG 1:"Methodology for En-
hancement and thematic Classification of visible and infrared
Data" of Commission VII
1035