Signature files generated by i.class

[Harini Nagendra]

Hello all,
I have classified a set of 20 landscapes using Grass4.1 for linux, using 
i.class followed by i.maxlike. After getting the classified landscapes, I 
have been trying to see how the landscapes look in terms of various 
landscape parameters.. patch sizes, shapes, landscape connectivity, etc. 
I want to now check how dependent the results of this landscape analysis 
are on the input training sets I have provided for i.class. What I 
thought is, I can play around with the signature files generated by 
i.class, change the means/std.deviations/covariances slightly, reclassify 
the landscapes with these, and then check how my landscape parameters 
look - this will give me an idea of how robust these parameters of 
landsacpe structure are. 
So - for this I need to know what each of the numbers in the signature 
files mean.. I have figured out part of it but need help on the rest. 
Also, any suggestions on the usefulness or otherwise of what I am 
planning to do would be appreciated.
I am enclosing a signature file below.. my imagery has 4 bands

#
#eucalyptus
16 - this is the number of pixels which I have marked
42.812500 23.875000 25.687500 61.187500 - mean intensities in Bands 1,2,3,4
0.429167                        | these 3 lines must be the covariances 
-0.091667 0.383333              | between bands - I need to know which 
0.270833 0.158333 0.629167      | one is for which pair of bands
-0.095833 -0.308333 -0.204167 0.829167 - this is the std.devns for bands 1234
 

I would really appreciate help on this.
Thanks a lot
Harini
------------------------------------------------------------------------
| Harini Nagendra               E-MAIL : harini at ces.iisc.ernet.in      |
| Center for Ecological         PHONE  : 91 80 309 2639 (Hostel: LR-94)|
| Sciences                             : 91 80 309 2506 (Department)   |
| Indian Institute of Science   FAX    : 91 80 334 1683                |
| Bangalore 560 012 India       TELEX  : 845 8349 IISc IN              |
------------------------------------------------------------------------