| Question on ENVI 3.6 [message #34612] |
Fri, 28 March 2003 05:45  |
Mort Canty
Messages: 134
Registered: March 2003
|
Senior Member |
|
|
Hi
The maximum likelihood classifier in ENVI version 3.5 generated a Rule File
with class membership probabilities. Version 3.6 generates something else.
According to the on-line help:
"The rule images, one per class, contain a maximum likelihood discriminant
function with a modified Chi Squared probability distribution"
This is not very informative and also contradicts the help elsewhere, where
it says:
"Use rule images to create intermediate classification image results before
final assignment of classes. For example, rule images for the maximum
likelihood classification will be the probability images themselves; one for
each input ROI used in the classification. The rule image that has the
highest probability for a pixel is the class to which that pixel is
assigned. The probability values themselves are only retained in the rule
images, not in the classified image. "
Can anyone tell me what the numbers in the Rule File actually are and how to
convert them back to probabilities?
Many thanks
Mort Canty
|
|
|
|
| Re: Question on ENVI 3.6 [message #34738 is a reply to message #34612] |
Sun, 13 April 2003 00:52  |
cshen
Messages: 6
Registered: April 2003
|
Junior Member |
|
|
hi, another question on max likelihood, but not quite realted to this
one.
When I used max likelihood, I use only input image, it is OK. But when
I use inpinoput image with a mask, in ENVI 3.5 and 3.6, in the
intermediate stage, an error message will appear (not in hand now).
Anyway meet similar situation? Thanks.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
