[gdal-dev] FillNodata and source NoDatas

[Jose Gomez-Dans]

Frank,

On 15 April 2010 17:24, Frank Warmerdam <warmerdam at pobox.com> wrote:

> Jose Gomez-Dans wrote:
>
>> Basically, my raster looks like a black square with some "island blobs"
>> where I have data. Some pixels in those blobs are flagged as unsuitable
>> (using a mask band), and I would like to interpolate these unsuitable
>> values. However, I want to make sure that the background NoData pixels in my
>> original raster will be ignored.
>>
>> Normally the raster fill algorithm uses the mask to identify pixels to
> be replaced, and it ignores the values of the pixels to be replaced.  Often
> the pixels to replaced are all nodata and everything else is valid.
>
> As I understand, the region you want to replace is a subset of all the
> nodata pixels, but you still want to ensure that nodata pixels which
> are not being replaced are not included in the interpolation.  Is that
> correct?
>

What you say is mostly correct. A simpler way to explain it (I ought to have
thought of that before!)  is that if you use a external mask dataset, the
raster fill algorithm should still honour ignoring the nodata values in the
original dataset, as these data are meaningless.

I can't entirely follow how the patch will work but it seems to be
> doing essentially a reasonable thing.   I am unfortunately hesitant to
> alter the GDALFillNodata() API at this point so I'm not sure your patch
> will
> make it back into the core. But once you have used it, I'd appreciate your
> submitting it as a ticket in case others need it.
>

I'll try it and see how it goes. Thanks for your help!
Jose
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/gdal-dev/attachments/20100415/27a7b810/attachment.html