| Re: frequency distribution and bytscl [message #31375] |
Wed, 10 July 2002 09:43 |
mchinand
Messages: 66
Registered: September 1996
|
Member |
|
|
In article <3D2C5C38.5F3C7E6A@ssec.wisc.edu>,
Liam E. Gumley <Liam.Gumley@ssec.wisc.edu> wrote:
>>
>> I think the PERCENT option of HIST_EQUAL function will do what you want.
>> It uses histogram and then TOTAL(histo,/CUMULATIVE) to find the cutoff
>> points.
>
> Mike,
>
> That's a very useful suggestion, and I'll certainly use it myself. This
> keyword does not seem to be available for HIST_EQUAL in IDL 5.3. Do you
> know if it became available in IDL 5.4 or 5.5?
>
I'm not sure. I'm running 5.5. The PERCENT keyword isn't mentioned in the
'What's New' files for either 5.4 or 5.5. It's described in the online
help, so it must have been added in 5.4 if it's not in 5.3.
--Mike
--
--
Michael Chinander University of Chicago
m-chinander@uchicago.edu (773)834-5101 (Voice) (773)702-0371 (Fax)
|
|
|
|
| Re: frequency distribution and bytscl [message #31376 is a reply to message #31375] |
Wed, 10 July 2002 09:09  |
Liam E. Gumley
Messages: 378
Registered: January 2000
|
Senior Member |
|
|
Mike Chinander wrote:
>
> In article <aghiu4$kcb$1@newsreader.wustl.edu>,
> Sean Raffuse <sean@me.wustl.edu> wrote:
>> Hello,
>>
>> I want to stretch an image based on it's frequency distribution:
>> stretched_image = bytscl(image, min=low, max=high)
>> where low is the bottom 5th percentile and high is the upper 5th percentile.
>> How can I get the values of the x percentile of a given array?
>> thanks in advance,
>>
>> Sean
>>
>
> I think the PERCENT option of HIST_EQUAL function will do what you want.
> It uses histogram and then TOTAL(histo,/CUMULATIVE) to find the cutoff
> points.
Mike,
That's a very useful suggestion, and I'll certainly use it myself. This
keyword does not seem to be available for HIST_EQUAL in IDL 5.3. Do you
know if it became available in IDL 5.4 or 5.5?
I wrote a short example function for my book that accomplishes the same
task. I've previously heard this technique referred to as "histogram
clipping". The name of the function is imclip.pro, and it's available in
the sample programs archive at
http://www.gumley.com/PIP/About_Book.html
Here's how it's used:
a = dist(256)
range = imclip(a, percent=5)
tv, bytscl(a, min=range[0], max=range[1])
or with my IMDISP program
imdisp, a, range=imclip(a, percent=5)
Cheers,
Liam.
Practical IDL Programming
http://www.gumley.com/
|
|
|
|
| Re: frequency distribution and bytscl [message #31377 is a reply to message #31376] |
Wed, 10 July 2002 08:44 |
mchinand
Messages: 66
Registered: September 1996
|
Member |
|
|
In article <aghiu4$kcb$1@newsreader.wustl.edu>,
Sean Raffuse <sean@me.wustl.edu> wrote:
> Hello,
>
> I want to stretch an image based on it's frequency distribution:
> stretched_image = bytscl(image, min=low, max=high)
> where low is the bottom 5th percentile and high is the upper 5th percentile.
> How can I get the values of the x percentile of a given array?
> thanks in advance,
>
> Sean
>
I think the PERCENT option of HIST_EQUAL function will do what you want.
It uses histogram and then TOTAL(histo,/CUMULATIVE) to find the cutoff
points.
Hope that helps,
--Mike
--
--
Michael Chinander University of Chicago
m-chinander@uchicago.edu (773)834-5101 (Voice) (773)702-0371 (Fax)
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
