ї . .
E-mail: fin_val_iv@tsure.ru.
347928, . , . , 44.
.: 88634371773.
Beloglazov Denis Aleksandrovich Bubley Sergey Evgenevich
Taganrog Institute of Technology - Federal State-Owned Educational Establishment of Higher Vocational Education Southern Federal University.
E-mail: fin_val_iv@tsure.ru.
44, Nekrasovskiy, Taganrog, 347928, Russia.
Phone: +78634371773.
004.932.2
..

- . . . .
; ; ; ; .
E.A. Vershovsky
QUALITY CONTROL METHOD FOR REMOTE SENSING DATA
CLUSTERING
Problem is considered about remote sensing data clustering results comparing. It is offered designed by author for her decisions method to clustering quality control. The broughted results of the comparison of the clustering quality control method with well known error matrix method. It is considered using of this problem in practice.
Remote sensing data; error matrix; accuracy matrix; clustering quality; control method.
- . , , ї - , [1]. , , , . Fisher's Iris data set [2,3]. .
, , .
, , , - .
:
;
.
.
, , , [4]. (. 1) ( ^) ( ).
1


A B Z I
A nAB nAZ -
B nBA nBB nBZ -
C nCB nCZ -
niZ -
Z nZA nZB nZC nZi nZZ nZ-
I -A - -c n-i n-Z N
, . . , ,
C B ,
nCB . :
n-A = + + + + + ... +nzA ;
- = + + + + nAi + ... +nAz; (1)
N = X n-i = X ni-.
( ).
, N - (Overall Accuracy, OAcc) :
OAcc = + + ncc + + + + + (2)
N '
. ї (Producers Accuracy, PAcc). A:
PAcc = ^ . (3)
(Ommission error, OErr). , ( ) (), . :
OErrA = 1 - PAccA . (4)
, . ї (User's Accuracy, UAcc) , . A:
n
UAcc = ^AA. (5)
(Commission error, CoErr). , , , . :
CoErrA = 1 - UAccA . (6)
[5], :
1. - . , A ї A . .
2. .
: , . -
n-A
(k-means, ISODATA), , . , , . , ( , , ..) (, Fuzzy c-means ISODATA).
(. 2). A1, 2,..., An - , , 1, 2... - , . , , ( , ..). , A1, 2,..., An 1, 2,..., . . . 2 , .
2


B1 B2 B3 Bn XAj
A1 cA1B1 cA1B2 cA1B3 cA1Bi CA1Bn CA1-
A2 cA2B1 cA2B2 cA2B3 cA2Bi CA2Bn CA2-
A3 cA3B1 cA3B2 cA3B3 cA3Bi CA3Bn CA3-
cAjB1 cAjB2 cAjB3 cAjBi CAjBn CAj-
An cAnB1 cAnB2 cAnB3 cAnBi CAnBn CAn-
C-B1 C-B2 c-B3 C-i C-Bn N
. 2 :
- = ,]=1---',
}- = 1, ,=1...; (7)
N = X. = , =1... , ]=1....
, :
1. ]=1.
2. = (]).
3. , . = () , . - - . /
4. ]= , , 1.
, .
dm - - , .
dAkBm , N - (Overall Accordance, OAcc) :
OAcc = ^ dABB . (8)
N
ї , . (Accordance Accuracy, AccAcc), , k- m- , , , .
A Ak, B, :
AccAccAk = dj4kBm . (9)
cAk -
B , Ak:
AccAccBm = dAkBm . (10)
c-Bm
(Accordance Error, AccErr) Ak A, , B, :
AccErrAk = 1 - AccAccAk . (11)
(11) . ,
(AkBm).
:
AccErr = 1 - AccAc^m . (12)
(Overall Accordance Accuracy, OvAccAcc) :
AccAccAi
OvAccAccA = ^=----------------. (13)
n
, (Overall Accordance Error, OvAccErr) :
yn AccAcc,
OvAccErrA ===-------------. (14)
n
. , , , / , .
, , .. . , , .
, .
- ( 10 () 100 ()). = 3.
(. 3).
.
3


1 2 3 ,
1 11 15556 16383 31950 51 49
2 97044 197 0 97241 100 0
3 17159 105429 0 122588 86 14
,, 114214 121182 16383 251779
85 87 100
15 13 0
=1 97044 +105429 +16383
=---- -----=------------------------= 87 %=
N 251779
. . (85 + 87 +100) (15 +13) .. /1
. =--------------= 91% = . ==----------------= 9%= 05)
3 3
= (51 + 10∞ + 86) = 79 %; == (49 +14) = 21%.
, 87 %. : 79 %, - 21 %, - 45,7. : 91 %, - 9 %, - 39,8.
, , , , . :
= \^1\, (I6)

j=k i=m, dAkBm = max(cAjBm)
dAkBm
= max(cAkBi),
OvErrB =
S=i OvErrB.
(17)
, 36 %.
. , , . , , (. 1,,), , , , . , (. 1,,).

. 1. : , , ,
, . [6].

1. Chulhee L.,Landgrebe D. Analyzing High Dimensional Multispectral Data // IEEE Transactions on Geoscience and Remote Sensing, 31. -1993. - 4. - . 792-800.
2. . . - .: , 2004. - 368 .
3. .. . - .: , 1982. - 216 .
4. [ ] // Gis-Lab: []. URL: http://gis-lab.info/qa/error-matrix.html ( 13.01.2010).
5. Glossary of remote sensing terms [ ] // The Canada Centre for Remote
Sensing []. URL: http://www.ccrs.nrcan.gc.ca/glossary/index_e.php?id=3124 (
13.01.2010).
6. .. // . . - 2010. - 5 (106). - . 102-107.
n

ї . .
E-mail: vershovsky@cbt.ru.
347928, . , . , 44.
.: 88634371673.
Vershovsky Eugeniy Alekseevich
Taganrog Institute of Technology - Federal State-Owned Educational Establisment of Higher Vocational Education Southern Federal Universityї.
E-mail: vershovsky@cbt.ru.
44, Nekrasovskiy, Taganrog, 347928, Russia.
Phone: +78634371673.
681.518
..

. , .
; ; ; .
Y.V. Klunnikova
MODEL OF THE SAPPHIRE GROWTH TECHNOLOGICAL PROCESS PARAMETERS INFLUENCE ON CRYSTALS QUALITY
The model of sapphire growth technological process parameters influence on crystals quality is presented in this article. The software and data ware for sapphire crystals growth process are developed. It allows to systematize the large information volumes and to give the exact crystals characteristics.
Information system; sapphire production technological process; quality, optimization.
( ), , . . . . [1].
. .