How to find the number of endmembers for other datasets?

[atheeraa]

Hello,

I'm trying to run your code with other datasets, but how do I find the estimated number of endmembers?
The actual number of endmembers is 4
I set c = 20 as to overestimate the number expecting that there will be redundant endmembers
from this I get:
W1Sum =
0.0062
0.0073
0.0083
0.0111
0.0131
0.0139
0.0144
0.0152
0.0162
0.0170
0.0175
0.0188
0.0200
0.0224
0.0225
0.0235
0.0255
0.0331
0.0376
0.0399

From what I understand from the paper is that the redundant endmembers weights should be almost zero, but in this case I cannot decide what threshold to set to separate the estimated endmembers from the redundant?
I'm trying to compare my thesis to your work and that's why I need to understand your code.
Thank you.