[gdal-dev] Erosive pyramids again

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[gdal-dev] Erosive pyramids again

Carlo A. Bertelli (Charta s.r.l.)
Hello, some time ago I complained with this list, asking for a better
algorithm for black and white and paletted raster maps.
While you scale the image, smaller black lines become thinner and
thinner, scaling to half size usually works without data loss but
further size reductions result into severe data loss.
I suggested introducting an "erose" option as to make thin lines
become fatter while reducing size.

Evaluating the consequences of the proposal, Even Rouault suggested
this was not a task for GDAL, but a user level choice to be provided
by QGis or other programs.
This is true, well known and expensive proprietary map-something show
antialised raster layers without any special effort, but at the
expense of huge computational requests.
Now, making pyramids should substantially reduce this burden, but
without severe data loss.

Obviously, Even is right, client programs should design a better
strategy to deal with line art and paletted maps, they could take into
account a near scale pyramid and an higher resolution one to draw
antialised layers in a slower rendition (this approach could be user
selectable or automatical, based on raster type), but it's still true
that GDAL should do its part not to lose data. And that happens
without any choice by the user than to avoid pyramids. Could we do
something better?
_______________________________________________
gdal-dev mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/gdal-dev
Reply | Threaded
Open this post in threaded view
|

[gdal-dev] Re: Erosive pyramids again

Rahkonen Jukka (Tike)
Carlo A. Bertelli (Charta s.r.l. <bertelli <at> charta.acme.com> writes:

>
> Hello, some time ago I complained with this list, asking for a better
> algorithm for black and white and paletted raster maps.
> While you scale the image, smaller black lines become thinner and
> thinner, scaling to half size usually works without data loss but
> further size reductions result into severe data loss.
> I suggested introducting an "erose" option as to make thin lines
> become fatter while reducing size.

Hi,

I would make a little study about JPEG2000 format. Wavelett compression could
well suit your use case because redused resolutions are built in a different
way. This page seems to have lots of sample images
http://preserve.harvard.edu/massdig/hul_study/ 

-Jukka Rahkonen-



_______________________________________________
gdal-dev mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/gdal-dev