b) Training area evaluation
The multi-spectral information of each training area vill be
displayed as shown on Fig.7. Trainees can check the
separability of each items at each spectral band. If the
separability of each items are not clear, trainees have to
select better training area again.
E 5;
se THRESHOLD ++ 3 127
-
>> CATEGORIES << BAND4 -
BI Forest
u 127
B Plant =
Wl suburban BANDS À -—
M Urban
9 127
M water ="
Bl Reclaimed BANDS | = -—
SEMI 127
whe
um
S T
Fig.7 Spectral Characteristics of Training Area
c) Classification
Based on the training data selected by the trainee, the multi
spectral classification(minimum distance classification) will
be performed. The trainee can compare his classification
result with the optimal classification result already prepared
by ENDIPS-T, and can repeat classification procedure. Through
this try and error process, the trainee can understand the
technique of classification.
4) Geometric Correction
Trainees are required to select GCP by comparing the ground
pattern of the two remotely sensed data taken at different time.
Geometric correction is automatically applied to the data by
using the GCP data. Trainees can compare and check the
registration accuracy of original image and geometric corrected
image by flickering the both image on the display.
5. Conclusion
As described above, ENDIPS-T allow beginners to understand basic
ideas of digital image processing and smoothly forward them to
the training of real remotely sensed data analysis. The
characteristics of ENDIPS-T can be summarized as follows.
50