ined 
or is 
tion 
Ly of 
es 
arror 
at to 
nflu- 
rices 
rob- 
xact 
ana- 
ance 
ssifi- 
July 
stin- 
and 
have 
Was 
lp of 
s for 
clas- 
re is 
n vi- 
map 
d for 
used 
10WN 
mple 
We see the great potential of object classifiers. The use 
of unsupervised classification can help to select training 
fields more accurately. It is planned to do in further ex- 
periments. The use of analytical methods for evaluation 
of the performance of classifier can help significantly to 
reduce computer time. This is also planned for further 
experiments. 
TABLE 1. The results of forest classification 
of nine classifiers 
  
  
  
  
  
  
Classifier | PR(%) | PK(%) 
PIX 76.22 75.99 
OMARK1 89.47 87.44 
OIND1 88.29 85.40 
OMEANIND1 80.81 80.14 
OMEAN1 67.91 67.91 
OMARK3 90.66 86.25 
OIND3 90.49 88.29 
OMEANIND3 82.68 82.34 
OMEAN3 68.93 68.93 
  
9. CONCLUSION 
In this work object classifiers incorporating spatial char- 
acteristics of an image during classification and based on 
Markov random field model are introduced. 
Two analytical methods for the evaluation of performance 
of object classifier based on calculating the probability of 
error are presented. 
First experimental results on forest classification, based on 
LANDSAT TM data, show the great potential of object 
classifiers comparing with per-pixel classifier. On the other 
hand we have to note the complexity of the problem and 
the need for further investigation. 
6. REFERENCES 
+ References from JOURNALS: 
Alfiorov, G.A., 1989. Contextual methods for the analysis 
of aerocosmic images. In: Advances in Science and Tech- 
nics, Series: Technical Cybernetics, Moscow, vol. 26, pp. 
105-136 ( in Russian ). 
Kalayeh, H.M. and Landgrebe, D.A., 1987. Stochastic 
model utilizing spectral and spatial characteristics. IEEE 
Trans. on Pattern Analysis »nd Machine Intelligence, 9:457- 
461. 
Ketting, R.L. and Landgrebe, D.A., 1976. Classification of 
multispectral image data by extraction and classification 
of homogeneous objects. IEEE Trans. Geosci. Electron- 
ics, 14:19-26. 
Landgrebe, D.A., 1980. The development of spectral- 
spatial classifier for earth observational data. Pattern 
Recognition, 12(3):165-175. 
Landgrebe, D.A., 1981. Analysis technology for land re- 
mote sensing. Proc. IEEE, 69(5):628-643. 
Mardia, K.V., 1984. Spatial discrimination and classifi- 
cation maps. Communications in Statistics. Theory and 
Methods, 13:2187-2197. 
487 
Palubinskas, G. 1988a. A comparative study of decision 
making algorithms in images modeled by Gaussian Markov 
random fields. International Journal of Pattern Recogni- 
tion and Artificial Intelligence, 2(4):621-639. 
Palubinskas, G. 1988b. Probability of misclassification 
of the Bayes decision rule for normal random vectors in 
the case of model and data inadequacy. In: S.Raudys, 
Ed., Statistical Problems of Control, Issue 82, Institute of 
Mathematics and Cybernetics Press, Vilnius, pp. 32-42 ( 
in Russian ). 
Palubinskas, G., 1990a. A review of spatial image recog- 
nition methods. In: Statistical Problems of Control, Issue 
93, Vilnius, pp. 194-214 ( in Russian ). 
Palubinskas, G., 1990b. Object classifiers in remote sens- 
ing. In: Statistical Problems of Control, Issue 93, Vilnius, 
pp. 215-231 ( in Russian ). 
Palubinskas, G., 1992. On approximation of the error of 
the Bayes decision rule. To appear. 
Swain, P.H., 1985. Advanced interpretation techniques for 
Earth data information systems. Proc. IEEE, 73(6):1031- 
1039. 
Switzer, P., 1980. Extensions of linear discriminant analy- 
sis for statistical classification of remotely sensed satellite 
imagery. Mathematical Geology, 12:367-376. 
+ References from BOOKS: 
Jensen J.R., 1986. Introductory Digital Image Processing: 
A Remote Sensing Perspective. Prentice-Hall, Englewood 
Cliffs, pp. 177-233. 
Fukunaga K., 1972. Introduction to Statistical Pattern 
Recognition. Academic Press, New York, pp. 50-67. 
Richards J.A., 1986. Remote Sensing Digital Image Anal- 
ysis: An Introductory. Springer-Verlag, Berlin, pp. 173- 
250. 
+ References from GREY LITERATURE: 
Guyon X., and Yao J.F., 1987. Analyse discriminante con- 
textuelle. In: Proc 5th International Symposium on Data 
Analysis and Informatics, Paris-France, Vol.1, pp. 43-52. 
Palubinskas, G., 1989. Object classifiers for textured im- 
ages. In: Mathematical Research, Vol.55 ( Proc. of the 
3rd International Conference CAIP’89 on Automatic Im- 
age Processing ” Computer Analysis of Images and Pat- 
terns ", Leipzig ( GDR ), September 8-10, 1989 ), eds. 
K. Voss, D.Chetverikov and G.Sommer, Akademie Verlag 
Berlin, pp.141-147. 
Palubinskas G., Cibas T., Repsys A., 1991. IMAX - im- 
age analysis and classification system. Technical Report, 
Vilnius, 9 pages. 
Pyka K., 1990. Data preprocessing influence on classifica- 
tion results ( in relation to forest classification ). In: Proc. 
Symposium of ISPRS Com. III in Wuhan, pp. 743-755. 
Schulz B.S., 1988. Hypothesis-free land use classifica- 
tion from LANDSAT-5 TM image data. In: International 
Archives of Photogrammetry and Remote Sensing, Kyoto, 
Vol. 27, pp. 736-742.