[gdal-dev] gdalwarp makes landcover files blocky when shrinking pixel size

[John Twilley]

Loïc, this sounds sorta like what I was doing in my code poorly, which
is why I turned to the command-line tools.  Do you know whether any of
the GDAL command-line tools can apply a majority filter with a large
moving window?

Jack.
--
mathuin at gmail dot com



On Fri, Jan 27, 2012 at 12:05, Dutrieux, Loïc <loic.dutrieux at wur.nl> wrote:
> Hi John,
>
> If I may add a bit to the discussion, applying resampling algorithm other than nearest neighbor to thematic layers such as land cover classification is not really recommended.
> However, if you want to smoothen the edges of your resampling output, I would suggest a two steps approach:
> - Resample using nearest neighbor.
> - Apply a majority filter with a large moving window.
>
> I cannot guarantee the result, but I guess it should work.
>
> Best regards,
>
> Loïc
>
> -----Original Message-----
> From: gdal-dev-bounces at lists.osgeo.org [mailto:gdal-dev-bounces at lists.osgeo.org] On Behalf Of John Twilley
> Sent: vendredi 27 janvier 2012 20:58
> To: Dmitry Baryshnikov
> Cc: gdal-dev at lists.osgeo.org
> Subject: Re: [gdal-dev] gdalwarp makes landcover files blocky when shrinking pixel size
>
> None of those resampling methods are well-suited to classified rasters as far as I can tell.
>
> The method that has worked best in the past was to take the majority of a number of the nearest neighbors.  The 'mode' algorithm in gdaladdo may do that, but I haven't checked the source code.
>
> Jack.
> --
> mathuin at gmail dot com
>
>
>
> 2012/1/27 Dmitry Baryshnikov <polimax at mail.ru>:
>> 27.01.2012 23:30, John Twilley пишет:
>>
>>> Cubic is fine for the elevation, but it doesn't work on the landcover.
>>> which is the problem I originally mentioned.
>>>
>>> Jack.
>>> --
>>> mathuin at gmail dot com
>>>
>>>
>>>
>>> 2012/1/27 Dmitry Baryshnikov<polimax at mail.ru>:
>>>>
>>>> 27.01.2012 20:19, John Twilley пишет:
>>>>
>>>>> After retrieving the files and building VRTs from them, I run the
>>>>> following gdalwarp commands:
>>>>>
>>>>> gdalwarp -q -multi -t_srs "+proj=aea +datum=NAD83 +lat_1=29.5
>>>>> +lat_2=45.5 +lat_0=23 +lon_0=-96 +x_0=0 +y_0=0 +units=m" -tr 6 6
>>>>> +-te
>>>>> 2002944 2271744 2016768 2288640 \
>>>>>  -srcnodata -340282346638528859811704183484516925440 -dstnodata 0
>>>>> ND302HT.vrt ND302HT.tif gdalwarp -q -multi -t_srs "+proj=aea
>>>>> +datum=NAD83 +lat_1=29.5
>>>>> +lat_2=45.5 +lat_0=23 +lon_0=-96 +x_0=0 +y_0=0 +units=m" -tr 6 6
>>>>> +-te
>>>>> 2002944 2271744 2016768 2288640 \
>>>>>  -srcnodata 255 -dstnodata 11 L0102HT.vrt L0102HT.tif
>>>>>
>>>>> The only differences between the lines are the nodata values.
>>>>>
>>>>> Jack.
>>>>> --
>>>>> mathuin at gmail dot com
>>>>>
>>>>>
>>>>>
>>>>> On Thu, Jan 26, 2012 at 22:49, Dmitry Baryshnikov<polimax at mail.ru>
>>>>>  wrote:
>>>>>>
>>>>>> 27.01.2012 3:39, John Twilley пишет:
>>>>>>>
>>>>>>> I am working with elevation and landcover data downloaded from
>>>>>>> the USGS.  I use gdalwarp to convert the data to a much smaller
>>>>>>> pixel. The elevation data works very nicely with cubic
>>>>>>> resampling, but the only resampling that works at all for the
>>>>>>> landcover data is nearest-neighbor and that's very blocky.  When
>>>>>>> I last worked with landcover data, I used a majority algorithm
>>>>>>> which produced smoother output -- but that algorithm is not implemented in gdalwarp.
>>>>>>> I am looking over the source to gdalwarp to see how hard it is to
>>>>>>> add a new algorithm.  Other than that, though, what options are
>>>>>>> available to me?  Thanks in advance!
>>>>>>> Jack.--
>>>>>>> mathuin at gmail dot com
>>>>>>> _______________________________________________
>>>>>>> gdal-dev mailing list
>>>>>>> gdal-dev at lists.osgeo.org
>>>>>>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>>>>>>>
>>>>>>>
>>>>>> Hi John,
>>>>>>
>>>>>> That is your command line string for gdalwarp?
>>>>>>
>>>>>> Best regards,
>>>>>> Dmitry
>>>>>>
>>>>>> _______________________________________________
>>>>>> gdal-dev mailing list
>>>>>> gdal-dev at lists.osgeo.org
>>>>>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>>>>
>>>> Hi John,
>>>>
>>>> Try to add "-r cubic" command parameter For example:
>>>>
>>>>
>>>> gdalwarp -q -multi -t_srs "+proj=aea +datum=NAD83 +lat_1=29.5
>>>> +lat_2=45.5 +lat_0=23 +lon_0=-96 +x_0=0 +y_0=0 +units=m" -tr 6 6 -r cubic
>>>> -te
>>>>
>>>> 2002944 2271744 2016768 2288640 \
>>>>  -srcnodata -340282346638528859811704183484516925440 -dstnodata 0
>>>> ND302HT.vrt ND302HT.tif
>>>>
>>>> You can found additional parameters description here:
>>>> http://www.gdal.org/gdalwarp.html
>>>> I have tested -r parameter and it works.
>>>>
>>>>
>>>> Best regards,
>>>> Dmitry
>>>>
>>>>
>>>> _______________________________________________
>>>> gdal-dev mailing list
>>>> gdal-dev at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>>>
>>>
>> Hi John,
>> You can use resampling methods:
>> near: nearest neighbour resampling (default, fastest algorithm, worst
>> interpolation quality).
>> bilinear: bilinear resampling.
>> cubic: cubic resampling.
>> cubicspline: cubic spline resampling.
>> lanczos: Lanczos windowed sinc resampling.
>>
>> And what method do you need?
>>
>>
>> Best regards,
>>    Dmitry
>> _______________________________________________
>> gdal-dev mailing list
>> gdal-dev at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>