Full text: Proceedings of the international symposium on remote sensing for observation and inventory of earth resources and the endangered environment (Volume 3)

  
    
  
        
  
   
   
     
   
   
  
  
  
  
  
  
   
   
   
  
   
   
    
  
  
  
  
91 100 STEP VARIABLE LANDSAT F VALUE 
44 100 NUMBER ENTERED BAND MONTH TO ENTER 
44 95 1 3 6 MAY 159.9462 
29 74 2 15 6 AUG 86.4440 
43 87 3 14 5 AUG 57.4195 
4 95 4 5 4. JUN 41.2215 
ul T 5 7 6 JUN 37.5246 
6 19 6 SEP 36.4801 
15 20 7 2 5 MY 31.4473 
10 95 8 9 4. JUL 19.1810 
51 90 9 11 6 JUL 19.6123 
50 83 10 18 5 SEP 12.9724 
75 100 11 6 5 JUN 12.0967 
12 4 7 MAY 7 1311 
59 100 13 13 à AUG 6.9163 
8 82 14 8 7 JUN 5.5010 
50 75 15 10 5 JUL 5.1027 
25 75 16 17 4 SEP 4.8474 
90 100 17 1 4 MAY 4.5549 
18 16 7 AUG 4.5532 
of training 19 20 7 SEP 2.9718 
; ; 20 12 7 JUL 2.2619 
analysis 
  
2 : : Table 2: Summary of stepwise linear discriminant analysis 
) 
00 
VOL 
D 
eb 
- 
VI. .Comparison.of classification results 
with ground truth data 
The comparison "classification results" with "ground truth" is displayed in 
Fig. 4: here, a capital letter represents perfect coincidence between both. 
White pixels represent areas where no ground truth was available. Minus-signs 
signify pixels for which ground truth and classification do not coincide. 
They appear mainly at the borders of two neighbouring fields due to the 
pixels beingspectrally mixed or due to misregistration in the overlaying of 
the channels or due to misdigitization of the ground truth information. 
  
A certain number of minus-signs arises from the letter reason: our process for 
digitization of the ground truth data and their superposition on the 
classification results has to be revised. The precision we have worked with 
is not yet adequate for the data we have. Therefore the example given here 
merely serves to demonstrate the method of pixel-wise comparison. 
  
round truth Apart from that we see some misclassified fields telling us that our trai- 
mn 2) ning sample set used for the discriminant analysis was not representative 
Minus-signs for those cases. 
   
	        
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.