N

(3)

 

L

N 

58 

...

 

12

3

L

N

NN

N







N



........................(1) 

1

L

i

i

i

L

st

i

NX

X

N











1

i

i

WX................................(3) 

N

i

i

(Weight)

 

st

X

i

i

N

W

N



 ....................................(2) 

N

i

i

i

X

i

N

(3)

st

X

i

X

W

i

i

N

i

i

 

(Linear Combination)

 

 

11

2

2

22

2

12

1

11

22

1

2

1

3

2

11

()

(

...

)

(

)

(

)...

(

)2

(

,

)2

(

,

()2

(,

)

st

ll

n

LL

L

ii

j

i

i

j

ii

ji

VX

VWX

WX

WX

WVX

WVX

WVX

WWCovXX

WWCovXX

WVX

WWCovXX





















3

)...



.......(4)

 

()

st

VX

2.3

 

 

2.3.1 

(Supervised Classification) 

()

,

ij

XX

Cov

i

i

X

j

j

X

i

X

j

X

()

,

ij

XX

Cov

0

(4)

 

2

()

1

()

L

st

i

i

VX

W







i

VX....................................(5) 

(Training Areas)

(

)

(

2001)

Richards and Jia (2006)

 

1. 

…

 

()

i

i

VX

(6) 

2. 

(Repre-

sentative Pixels)

(Training 

Areas) 

2

()

i

ii

i

ii

sNn
nN



2

i

S

VX



 ....................................(6) 

i

n

i

i

(6)

(5)

 

3. 

(Signature) 

2

2

2

2

1

1

()

L

L

ii

i

i

st

i

i

i

i

ii

sN

n

s

VX

W

W

nN

n











(1

)

i

i

f



 

4. 

 

 ....................................(7) 

W

i

i

f

i

i

 (f

i

 = 

n

i 

/N

i

)(1 

 f

i 

)

f

i

5%

(Cochran, 

1977) 

5. 

 

6. 

 

2.3.2 

(Bayes’ Cslassification )