| Re: Surface fitting image with missing data [message #61804] |
Thu, 31 July 2008 01:27  |
werner.hupfer
Messages: 1
Registered: July 2008
|
Junior Member |
|
|
On 31 Jul., 02:12, jradavenp...@gmail.com wrote:
> Hey all;
>
> I have a 2-D array (essentially an image) with dimensions [n,m]
>
> There is a large chunk in a corner of the image with no data. All
> pixels have values greater than 1, except within this chunk, which =
> 0.
>
> SFIT fits these as valid points and tips my surface (a plane) towards
> this corner! I thought I would be terribly clever by doing the
> following:
>
> result = sfit( finite(image/(image gt 0)),1,/max_degree)
>
> alas, SFIT isnt playing ball and makes a whole array of /NaN...
>
> Thoughts? Is there a version of SFIT (or something similar) with a /
> NaN keyword? A more clever way to fit a simple surface to an image
> with missing data? Thanks folks
>
> - Jim
Hello Jim
I can highly recommend the library from Craig Markwardt. I use
mpfitfun for 2D fitting. Among one of the many useful features is the
ability to do curvefit on data points set to NaN!!!!
http://cow.physics.wisc.edu/~craigm/idl/idl.html
Cheers
Werner
|
|
|
|
| Re: Surface fitting image with missing data [message #61902 is a reply to message #61804] |
Sun, 03 August 2008 21:52  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
werner.hupfer@googlemail.com writes:
> On 31 Jul., 02:12, jradavenp...@gmail.com wrote:
>> Hey all;
>>
>> I have a 2-D array (essentially an image) with dimensions [n,m]
>>
>> There is a large chunk in a corner of the image with no data. All
>> pixels have values greater than 1, except within this chunk, which =
>> 0.
...
>
> I can highly recommend the library from Craig Markwardt. I use
> mpfitfun for 2D fitting. Among one of the many useful features is the
> ability to do curvefit on data points set to NaN!!!!
>
> http://cow.physics.wisc.edu/~craigm/idl/idl.html
Thanks for the plug!
Actually MPFIT doesn't accept NaNs but probably should. Even so, you
can set the WEIGHTS to zero for your missing data (or equivalently,
the ERRORs to a very large number).
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@REMOVEcow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
