[GRASS-dev] bilinear and bicubic raster interpolation

[Markus Neteler]

On Sun, Oct 17, 2010 at 6:56 PM, Glynn Clements
<glynn at gclements.plus.com> wrote:
>
> Markus Neteler wrote:
>
>> >> However, there is another issue with r.proj (7.0) and r.proj.seg
>> >> (6.x): bilinear and cubic interpolation give a small number of garbage
>> >> results due to incorrect use of the caching mechanism (blocks are
>> >> being replaced while references remain). I'll fix this shortly.
>> >
>> > This should be fixed by r43944.
>>
>> The common backport question comes up for this one and r43943...
>> Indications welcome.
>
> Yes to both. Note that 7.0's r.proj is r.proj.seg in 6.x, which may
> complicate matters.

Slightly... ok I have done both as

G6.4:
http://trac.osgeo.org/grass/changeset/43953
http://trac.osgeo.org/grass/changeset/43956

G6.5:
http://trac.osgeo.org/grass/changeset/43954
http://trac.osgeo.org/grass/changeset/43955

Please verify.

> Also, I believe that the off-by-half issue (r43943) affects the 6.x
> r.proj, which will need a similar fix (but which will need to be done
> manually). r43944 is only applicable to the newer version (r.proj.seg
> and 7.0), not the 6.x r.proj.

AFAIK we have r.proj.seg everywhere.

> FWIW, I found the errors using:
>
> r.plane --o name=plane dip=30 azimuth=45 easting=599505 northing=4921010 elevation=1453 type=double
> r.proj --o --q in=plane location=spearfish57 mapset=glynn output=plane.n method=nearest
> r.proj --o --q in=plane location=spearfish57 mapset=glynn output=plane.l method=bilinear
> r.proj --o --q in=plane location=spearfish57 mapset=glynn output=plane.c method=cubic
> r.mapcalc --o 'diff.n = plane.n - plane'
> r.mapcalc --o 'diff.l = plane.l - plane'
> r.mapcalc --o 'diff.c = plane.c - plane'
> r.info -r diff.n
> r.info -r diff.l
> r.info -r diff.c
>
> [I had to modify r.proj not to complain about the source and
> destination being identical. There's no reason why this should cause
> problems, although I can't see any use for it other than testing.]

Markus