First of all. providing the annual ground truth data is 
  
E time-consuming and involves some expense. However, 
| it might be managed in relevant student fieldwork with 
5m guide of experienced people, as ITC does. 
/ 0. 
/ om The second problem is to obtain up-to-date aerial 
" photographs of area. Having the regular aerial photos is 
too expensive. There is possibility to use old 
cutter photographs in areas with no great change in parcel 
kelihood shapes. 
The third problem is access to satellite imagery. 
mproved Normally, there are some official organizations or 
accuracy companies in charge of presenting satellite images. 
ventional However, there must be the possibility to order the 
images in growing seasons of vegetation. 
Having an accurate DEM is the most important thing 
among the others. The more the accurate DEM, the 
showing 
" more accurate the geometric correction process we can 
Van do. In this case study, we used the DEM that had been 
m compiled from digitizing and interpolation of 1:25000 
: map contours. It seems that this DEM does not have 
ix adequate accuracy for performing the necessary 
corrections. This lack of accuracy in DEM causes 
he DEM unmatched respective polygons and parcels in the 
ould not overlaid images. Unmatched polygons and parcel, then 
metry of cause uncertainty in classification of unages. 
ideration 
mproved 
es References 
picted in Richards, John A., 1993, Remote Sensing Digital Image 
Analysis, Springer-Verlag, Berlin. 
Sr da Russ. John C.. 1992. The [mage Processing Handbook, 
ed etrors M 
Mii CRC, Florida. 
ies were a LR UE TE ee 
Axe relief Ellis, E.C. Statistical Pattern Recognition - Classification 
and Feature Selection in Image Processing, ITC Lecture 
note, The Netherlands. 
gned the 
at parcel. Huumeman, G.C. Geometric Corrections of Digital 
n expect images, [TC Lecture note, The Netherlands. 
Kostwinder, H.R. and Bakx, J.P.G., 1991. Radiometric 
Corrections of Digital Images, ITC Lecture note, The 
Netherlands. 
case the = 
acy and vore, B.G.H. Local Operators/Filters and Structural 
lata have Pattern Recognition, ITC Lecture note, The Netherlands. 
dated or 
anges of Bak, IPG, 1991. Colour Coding and spectral Feature 
roblems. Extraction, ITC Lecture note, The Netherlands. 
a, aerial 
me (or at Horn, J., 1992 Primarv Data Acquisition, ITC Lecture 
f course, note, The Netherlands. 
t of data. 
541 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996