| Re: interpolation for resizing [message #64136] |
Wed, 03 December 2008 13:19  |
pgrigis
Messages: 436
Registered: September 2007
|
Senior Member |
|
|
I really love this ;-)
IDL> a=float('nanny')
IDL> help,a
A FLOAT = NaN
Chris wrote:
> On Dec 3, 6:36�am, Paolo <pgri...@gmail.com> wrote:
>> David Fanning wrote:
>>> bryan.s.h...@gmail.com writes:
>>
>>>> Only works in a real array not in a integer array.
>
> It also only seems to work for strings which correspond to actual
> floating point values, like 'nan' and 'inf':
>
> IDL> a = fltarr(3)
> IDL> a[0] = 'nan'
> IDL> a[1] = 'hi'
> % Type conversion error: Unable to convert given STRING to Float.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Re: interpolation for resizing [message #64148 is a reply to message #64147] |
Wed, 03 December 2008 07:49  |
bryan.s.hong
Messages: 12
Registered: November 2008
|
Junior Member |
|
|
On 12월3일, 오전7시52분, loebasboy <stijn....@gmail.com> wrote:
> On 3 dec, 14:43, David Fanning <n...@dfanning.com> wrote:
>
>
>
>
>
>> Craig Markwardt writes:
>>> That's true, but nearest-neighbor sampling also adds high-frequency
>>> aliases, so what one uses to interpolate does depend on what one
>>> needs.
>
>> I have a feeling it will be some weeks before the person
>> asking the question feels compelled to ask about this
>> complication. ;-)
>
>> Cheers,
>
>> David
>
>> P.S. Isn't a 100 by 100 image, uh, pretty small to be
>> doing *any* high-resolution analysis? Certainly too
>> small to be worried about high-frequency aliasing.
>> You should be thinking about other problems, my friend!
>
>> --
>> David Fanning, Ph.D.
>> Fanning Software Consulting, Inc.
>> Coyote's Guide to IDL Programming:http://www.dfanning.com/
>> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
>
> About interpolate and other related functions... is it possible to use
> these function to interpolate data where there is no data in a grid.
> For instance I have the following code:
>
> a= findgen(5,5)
> a[3,3] = -999
> a[1,2] = -999
> indices = where(a EQ -999)
> b = interpolate(a, indices)
>
> and the interpolated numbers in b are interpolated with available
> numbers in 2 dimensions...
>
> Is there a function in IDL that can do this, because interpolate and
> other related functions work differently I guess (making extra rows or
> columns between rows or columns). I'm sorry that this is a very 'open'
> question but I'm rather busy and this something I could use later on.- 따온 텍스트 숨기기 -
>
> - 따온 텍스트 보기 -
Replacing the fill value to a string, for example 'NaN', will exclude
the pixels when using 'congrid' or 'rebin'.
|
|
|
|
| Re: interpolation for resizing [message #64156 is a reply to message #64148] |
Wed, 03 December 2008 05:52 |
loebasboy
Messages: 26
Registered: August 2008
|
Junior Member |
|
|
On 3 dec, 14:43, David Fanning <n...@dfanning.com> wrote:
> Craig Markwardt writes:
>> That's true, but nearest-neighbor sampling also adds high-frequency
>> aliases, so what one uses to interpolate does depend on what one
>> needs.
>
> I have a feeling it will be some weeks before the person
> asking the question feels compelled to ask about this
> complication. ;-)
>
> Cheers,
>
> David
>
> P.S. Isn't a 100 by 100 image, uh, pretty small to be
> doing *any* high-resolution analysis? Certainly too
> small to be worried about high-frequency aliasing.
> You should be thinking about other problems, my friend!
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
About interpolate and other related functions... is it possible to use
these function to interpolate data where there is no data in a grid.
For instance I have the following code:
a= findgen(5,5)
a[3,3] = -999
a[1,2] = -999
indices = where(a EQ -999)
b = interpolate(a, indices)
and the interpolated numbers in b are interpolated with available
numbers in 2 dimensions...
Is there a function in IDL that can do this, because interpolate and
other related functions work differently I guess (making extra rows or
columns between rows or columns). I'm sorry that this is a very 'open'
question but I'm rather busy and this something I could use later on.
|
|
|
|
|
|
|
|
|
|
| Re: interpolation for resizing [message #64163 is a reply to message #64162] |
Tue, 02 December 2008 20:04 |
Chris[6]
Messages: 84
Registered: July 2008
|
Member |
|
|
On Dec 2, 1:28 pm, David Fanning <n...@dfanning.com> wrote:
> bryan.s.h...@gmail.com writes:
>> I have a 100 X 100 spatial image to be resized to 1000*1000.
>> The data of this image is surface temperature and includes some part
>> of ocean.
>> Because I want to exclude data of ocean area during interpolation, I'm
>> trying to resize it using "interpolate".
>> But I cannot understand the IDL help pages for the 'interpolate'.
>> Could anyone help me with a simple command line for this job?
>
> I don't think this is what you want to do.
> Just use CONGRID (or REBIN) to resize your
> image. (CONGRID uses INTERPOLATE to do the
> interpolation, so you can see how it is done,
> if you read the code.) Then mask out your ocean
> pixels in the usual way, with the WHERE function.
>
> bigImage = Congrid(image, 1000, 1000)
> oceanPixels = Where(bigimage EQ 9999, count)
> IF count GT 0 THEN image[oceanPixels] = 0
>
> Cheers,
>
> David
> --
> David Fanning, Ph.D.
> Coyote's Guide to IDL Programming (www.dfanning.com)
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
One potential issue with this approach is that IDL may interpolate
between your 9999 values and your real data values when you expand the
array. If you set the 9999s to !values.f_nan beforehand, then the
interpolation will behave more sensibly.
chris
|
|
|
|
| Re: interpolation for resizing [message #64166 is a reply to message #64163] |
Tue, 02 December 2008 15:28 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
bryan.s.hong@gmail.com writes:
> I have a 100 X 100 spatial image to be resized to 1000*1000.
> The data of this image is surface temperature and includes some part
> of ocean.
> Because I want to exclude data of ocean area during interpolation, I'm
> trying to resize it using "interpolate".
> But I cannot understand the IDL help pages for the 'interpolate'.
> Could anyone help me with a simple command line for this job?
I don't think this is what you want to do.
Just use CONGRID (or REBIN) to resize your
image. (CONGRID uses INTERPOLATE to do the
interpolation, so you can see how it is done,
if you read the code.) Then mask out your ocean
pixels in the usual way, with the WHERE function.
bigImage = Congrid(image, 1000, 1000)
oceanPixels = Where(bigimage EQ 9999, count)
IF count GT 0 THEN image[oceanPixels] = 0
Cheers,
David
--
David Fanning, Ph.D.
Coyote's Guide to IDL Programming (www.dfanning.com)
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: interpolation for resizing [message #64211 is a reply to message #64148] |
Fri, 05 December 2008 00:36 |
loebasboy
Messages: 26
Registered: August 2008
|
Junior Member |
|
|
On 3 dec, 16:49, bryan.s.h...@gmail.com wrote:
> On 12월3일, 오전7시52분, loebasboy <stijn....@gmail.com> wrote:
>
>
>
>
>
>> On 3 dec, 14:43, David Fanning <n...@dfanning.com> wrote:
>
>>> Craig Markwardt writes:
>>>> That's true, but nearest-neighbor sampling also adds high-frequency
>>>> aliases, so what one uses to interpolate does depend on what one
>>>> needs.
>
>>> I have a feeling it will be some weeks before the person
>>> asking the question feels compelled to ask about this
>>> complication. ;-)
>
>>> Cheers,
>
>>> David
>
>>> P.S. Isn't a 100 by 100 image, uh, pretty small to be
>>> doing *any* high-resolution analysis? Certainly too
>>> small to be worried about high-frequency aliasing.
>>> You should be thinking about other problems, my friend!
>
>>> --
>>> David Fanning, Ph.D.
>>> Fanning Software Consulting, Inc.
>>> Coyote's Guide to IDL Programming:http://www.dfanning.com/
>>> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
>
>> About interpolate and other related functions... is it possible to use
>> these function to interpolate data where there is no data in a grid.
>> For instance I have the following code:
>
>> a= findgen(5,5)
>> a[3,3] = -999
>> a[1,2] = -999
>> indices = where(a EQ -999)
>> b = interpolate(a, indices)
>
>> and the interpolated numbers in b are interpolated with available
>> numbers in 2 dimensions...
>
>> Is there a function in IDL that can do this, because interpolate and
>> other related functions work differently I guess (making extra rows or
>> columns between rows or columns). I'm sorry that this is a very 'open'
>> question but I'm rather busy and this something I could use later on.- 따온 텍스트 숨기기 -
>
>> - 따온 텍스트 보기 -
>
> Replacing the fill value to a string, for example 'NaN', will exclude
> the pixels when using 'congrid' or 'rebin'.- Tekst uit oorspronkelijk bericht niet weergeven -
>
> - Tekst uit oorspronkelijk bericht weergeven -
I guess you didn't read my question properly (although I maybe didn't
have to ask it in this thread to avoid this, but I didn't want to make
a new thread because of David's post several days ago). I'm not trying
to resize anything. I just want the -999 or NaN (I make them with !
Values.F_NAN option btw.) values to have a value in the b array that
is calculated by interpolation (for instance: the IDW of the
surroundig pixels in a 3x3 window).
So sorry for the mixup and thanks if you guys still feel like finding
a solution for this.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
