Full text: Proceedings, XXth congress (Part 8)

International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 Int 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
T 
Figure 2. Test area was classified by supervised classification (ENVI) 
Comparing the results of the various classification methods between the calculated and the real landuse categories (table 
the maximum likelihood one provided the least difference 1). 
Table 1. Confusion matrix for maximum likelihood classification ; 
Class name Reference total | Classifed Correctly | Accuracy 
pixel total pixel classified % 
alfalfa 52800 58429 52100 98,67 
maize (1,2,) 69200 53900 77.22 
maize (1) 44380 - 38024 85,67 
maize (2) 24740 - 15338 61,99 
sugar beet 43980 43620 42894 97.53 - 
wheat stubble 8395 18650 8237 98,11 
bare field 15605 15301 15143 97,03 
Total 189900 189900 171736 90,43 
(maize 1.: low number of crops, weedy maize, maize 2.: normal number of crops) 
Considering the spectral profiles of the certain plants we were images were taken to detected the changing of the land use. 
looking for the wavelength range, where there were the With help of the image complied by supervised classification 
largest distances gained between the studied values. and the layer containing buffer zones related to the 
Comparing the variance coefficients the most differences in prescriptions fields can be defined, which are not cultivated Al 
spectrums from 623 to 657nm (DAIS bands 8-10) and from as it had been negotiated. A 
814 to 904nm (DAIS bands 19-24) were observed, but if only DISCUSSION AND CONCLUSION ag 
the vegetation covered regions were taken into consideration 
the range from 762 to 886nm (DAIS bands 16-23) showed - Using the supervised classification we could gain C: 
the biggest variance coefficient (CV=0,30-0,33). accurate landuse results only in case of the Fr 
homogeneous plant cover. Through the decreasing of ve 
Whilst we the teaching areas pointed out for study of various vegetation cover the reflection of the soil increases 3 
vegetations and landuse types were being analysed we found, strongly modifying the reflectance values. 
that in the crop with canopy over 75% the obtained - . The island-like protected associations with a small Di 
reflectance values remained homogeneous (CV- 0,008- expanse and the breeding habitats can be classified well lai 
0,027) showing no significant correlation to the local soil and separated from cropped area using these large Jo 
facilities. Lowest diversity were gained at the longer spatial resoluted images. Fr 
wavelengths. In the case of disturbed crop with lower number - Images taken in various period during the year are lai 
of plant stocks the bigger values of variance coefficient were practical for to complete the sate assessment and the Er 
caused by the differences in reflection of the soil and the control functions (table 2). Through the combination of 
green leafs. images taken in different seasons exact landuse map can Fl 
be complied. In 
We made additional images from air and ground with a 
TETRACAM ADC wide band multispectral camera. Several L. 
> 19 
La 
Re 
Mi 
rel 
me 
31 
88
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.