vity and 
ctory to 
made an 
"Wards a 
ntextual 
thematic 
allowed 
condary 
for the 
issue for 
n cover 
lly more 
ding the 
a longer 
adequate 
ification 
network 
(92%) 
> classes 
and the 
| use of 
mplexity 
1993. À 
Landsat 
rt CCR- 
al para 
nazónica 
Thesis, 
123p. 
e lógica 
artificial 
 Tucurui 
Pesquisas 
| 16p. 
orace, F. 
based on 
ksi 
Sensing 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Segmented 
TM/Landsat 
Image 
+ 
Labelling of 
Segments 
i 
= °°... RTT a 
| l 
I Mean —+ Variance |— Correlation | —L- Entropy I 
| | 
EL —————— étre — e! 
  
  
  
    
   
  
  
Performance 
evaluation 
  
  
  
  
  
  
Classification 
  
  
  
  
  
  
  
  
  
  
  
  
  
i entral Network 
p — ET 
| : l 
| Mean || Variance 1] Correlation || Entropy |HNeighborhood | 
| | 
  
  
Thematic Map 
Figure 1 - Flow diagram of the neural network approach. 
  
  
    
  
  
TM3 TM 4 TM5 
--- Azi 
--- Azl A75; 17.25 
A5, 1-25 A- 75; F - 25 
— A=.75; F=.25 
Frequency 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
PEUT IUE IT TnL IE ree 
0 4 8 12 16 20 24 28 3 36 40 4 48 5 o 20 40 eo so 100 120 140 0 10 20 30 40 50 60 70 80 90 100 110 120 
Grey Level of Segments Grey Level of Segments Grey Level of Segments 
TM3 TM4 TMS 
CM RAS 
--- Jel --- Ip > 
1=.75 A=25 LN 
I7.75; A7 25 Heer 
—— | = 75; O= 25 
— 15.75, 0- 25 | \ 
\ 
| \ 
p I 
5 I \ 
g g 7 ' I 
= u. FUN = ; : 
; \ : \ 
\ | 
/ 1 
\ 
/ t. 1 \ 
/ \ ; 
/ \ 
> Y / 
l LT T1 LT. 1 wry MEI EG. um dom M S ME EM T I E zheddaL L1. 1p l| |^ 
0 4 8 12 10 20 24 28 32 36 40 44 48 52 50 60 0 20 40 eo so 100 120 140 © 10 20 30 40 50 60 70 80 90 100 110 120 130 140 
Grey Level of Segments Grey Level of Segments Grey Level of Segments 
Figure 2 - Frequency of spectral descriptors for "initial and advanced 
secondary succession" (Santos et al., 1995). 
207 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996