EEE ESTG
ENT —
date:1976
code | size | 15th | 19th | 17th 3rd 18th temp.
category
lake - See S 45 | 100 97 97 97 91 100
reed - Schilf C 18 44 27 50 27 44 100
4oods- Wald W 98 65 78 61 86 44 95
urban - Dorf/Weiler D 89 22 28 11 19 29 74
barley - Gerste G 30 20 56 33 16 43 87
winter-wheat - Winterweizen| T 19 47 52 37 10 47 95
summer-wheat - Sommerweizen | Z 29 65 31 31 31 24 80
rye - Roggen 0 40 15 45 10 55 15 80
potato - Kartoffel K 19 57 52 0 0 10 95
sugar-beet - Zuckerrueben | U 29 72 41 24 45 51 90
rape-seed - Ruebsen B 12 50 50 33 50 50 83
rape - Raps A 12 83 91 33 75 75 100
meadow-land - Wiese/Weide | R 32 47 28 25 81 59 100
corn - Mais I 39 23 44 62 71 8 82
pea - Erbsen E 75 50 100 75 50 75
vegetables - Gemuesebau M 4 50 0 75 50 25 75
vine-yards - Reben N 10 | 100 90 90 80 90 100
A EE
zzz
PT
DO
oe SER on RC SARE WRT
ZZZZIIZIIz2z2
2222222222
e$225222757
:
T
€——— N
ss
zz
nj
pe
Q d
——Ó—
Table 1: Percentages of correctly classified pixels of training
samples calculated by linear discriminant analysis
U---DUUDDUDDD--
07 D- D-- DuDDU--
DULL = Dm mem
-1 UD" DL
future DODDUDDDDDD- DUDOD
-$--——-6-- 000-—— DO- DDDDDDDDDDDDDDD
Ba DELETE foe
T IPHIIIHIIIIII
i ° RRRk==RRKRARRARAKKKKP
»- ~~111 Bbb-—- — RRKRRR-RRRRRRRRRRRKRE, -
—---RRRR---RR R--BB--——— RRRRRRRRRR- RKRRRRRRAKnnannR ERA
$55555-C-WwhW--- -DUDDODDD- RKRRRKRRR-TTTTTT-AA--KKRRK-TTTTT-GGGGRA- anna ne aM me
S. Hm
S5555$5-CL.
SE Tm
111 TTTT-
1 ~TTTT
n TD te —
si -5
11
1111
cc mm
——— —- OUUU00000-GOO- 11 1-AAA-- 11 IR^ 1111788817 RR —
————--0000000--0--1111----111-RkAR- 11 1 1--EE--K
ccc Wuntmi- --UU0- Qe 1E Le KR RR L-RRRRRR- 1177 E777
sod
CCCCC--CCRmehmhR--ÜOU- — ----- U-UUU---1133d------- do
c: Wm ROO == -T-- ^ U--U--1--1-00--- wa in
CnC = Al Yom B Emenee man ba
imine 1 RU GG RR Deren nieto P aisi ipo
— BRER-- P
— WW c — - XKRRKK WM in a
ee iem x
Cu in BE vor
e ———l rouen mee
— 1 yide
wm 1- MO irit
— WE Mene
3 An
135 —
— musee
a.
~ aie
Mette M
"—
1 1 1 1 1 1 1 1 1 1 1 1 1
4: Comparison of classification results with ground truth
data (code for categories see Table 1, column 2)
coincidence: capitals - no coincidence: Minus-signs
white areas: no ground truth available
Tabl
JI Co
Ww i
The compa
Fig. 4: h
White pix
signify p
They appe
pixels be
the chann
A certain
digitizat
classific
is not ye
merely se
Apart frc
ning samp
for those
