comp.lang.idl-pvwave archive: archive » class_doit Maximum Likelihood errors

Home » Public Forums » archive » class_doit Maximum Likelihood errors

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

class_doit Maximum Likelihood errors [message #50338]

Sun, 24 September 2006 20:13

humphreymurray
Messages: 8
Registered: August 2006

Junior Member

Hi,

I want to classify an 8 band image using the maximum likelihood
classifier in ENVI 4.0. I want to use a script to automate this
process, as I want to do it over a large range of different images and
settings.

However, if I classify all eight bands using the ENVI GUI I get a
completely different classification image then I do when I do it in
code. If I only use four or less of the bands, then the results are
exactly the same, however if I try to use anymore then four of the
bands the results using my code as really bad, and look nothing like
the results produced by using the ENVI GUI.

Am I doing something wrong here? Or is this a known (or unknown) bug.

I even tried copy and pasting the code from the class_doit manual page
(and pointed it to my image), and it still produced a wrong result.

Humphrey

Report message to a moderator

Re: class_doit Maximum Likelihood errors [message #50490 is a reply to message #50338]

Thu, 28 September 2006 11:24

peter.eddy@shaw.ca
Messages: 19
Registered: March 2005

Junior Member

It is difficult to troubleshoot the problem without seeing the code you
are trying to run. First I would go back and check your training data
file. The ML classification as im sure you know uses training data's
variance / covariance matricies... Are your *.sta files or matricies
present and formatted properly?? Were your training matricies developed
on only four bands?

Pete

Report message to a moderator

Previous Topic:	Expanding Arrays of Structures
Next Topic:	IDL on Ubuntu?

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Wed Oct 08 19:24:16 PDT 2025

Total time taken to generate the page: 0.00541 seconds