CLUSTERING OF MULTISPECTRAL DATA 
BASED ON GENERALIZED DISTANCE CRITERION FUNCTION 
by 
S. CHANDRASEKHAR 
M. MARUTMACHALAM 
and 
K.R. RAO 
National Remote Sensing Agency 
No.4, Sardar Patel Road 
Post Box No. 1519 
Secunderabad - 500 003. A.P. 
INDIA 
This paper briefly discusses a clustering technique on 
generalized distance criterion function optimization. 
The criterion function used is a quadratic function 
based on total scatter matrix and distance of samples 
from the means.  Initially, all the sample points to be 
clustered from individual cluster centres and subsequen- 
tly clusters are merged depending upon change in cri- 
terion function till the desired number of clusters are 
obtained. The algorithm was tested on Landsat data and 
it was found to perform quite good when compared to 
least square criterion function. The results are 
discussed, 
79 
  
| 
| 
| 
|