06 Jun 2008 11:38 AM |
|
When I read about the Jeffries-Matusita distance I find info that it ranges from 0 - 2000 or 0 to the sqrt of 2, yet in ENVI the values range from 0 to 2.0. Can someone explain this?
Thanks!
|
|
|
|
Deleted User New Member
Posts:  
06 Jun 2008 11:38 AM |
|
Thanks for the info.
Can you also tell me if it is assuming the classes are normally distrbuted? It seems like you might be able to work the equation in 2 ways depending on if the data is / is not normal.
Thanks again
|
|
|
|
MariM Veteran Member
Posts:2396  
06 Jun 2008 11:38 AM |
|
It is the Transformed Divergence that is defined between the ranges of 0-2000. The J-M Distance is defined as the sqrt of 2(1 - e ^-alpha) where alpha is the bhattacharya distance.
The Jeffries-Matusita distance has an upper bound of 1.41 * sqrt(2) and a lower bound of 0.
I hope this helps!
|
|
|
|
Deleted User New Member
Posts:  
29 Jul 2012 05:59 PM |
|
But ENVI shows the J-M distance squared... so the upper bound is actually increased to 2.
|
|
|
|