[GRASSLIST:126] Re: making a "true-color" landsat composite

[Markus Neteler]

Glad to hear that i.landsat.rgb works well.

I sort of hesitate to add it to CVS since it could be
"easily" added to r.colors.

I cc to grass-dev to seek for a solution. The only
real missing function in the C library is the possibility
to calculate percentiles from a channel histogram.

Wolf was working on that (percentiles in r.univar) but
apparently it's not complete. The approach could be:

- implement percentiles and related stats in C (gmath?)
- add percentiles to r.univar
- add percentiles to v.univar
- remove r.univar.sh as no longer needed
- remove v.univar.sh as no longer needed
- implement method in r.colors (the LANDSAT color trick is
  to take away 2% of the histogram area from top and bottom)

Having this long term wish "percentiles" implemented in C
would in fact solve a couple of problems (duplicated modules).

I really hope that we can find another capable to help Wolf
(if he's still available).

Markus

On Wed, Mar 15, 2006 at 12:12:23AM -0800, Dylan Beaudette wrote:
> Thanks again Markus:
> 
> your new i.landsat.rgb script did wonders on these images!
> 
> examples:
> 
> original made with r.composite:
> http://casoilresource.lawr.ucdavis.edu/drupal/node/217
> 
> new one made with i.landsat.rgb:
> http://casoilresource.lawr.ucdavis.edu/drupal/node/216
> 
> Cheers!
> 
> Dylan
> 
> On 3/14/06, Dylan Beaudette <dylan.beaudette at gmail.com> wrote:
> > Ian,
> >
> > Thanks for the pointers. In the past, I was not performing any
> > "enhancements"
> > to the landsat scenes as there were so many of them, and they looked good
> > enough. However with Markus' latest script, I think that I will re-compile
> > arizona! good thing that I hadn't started on Nevada yet!
> >
> > Cheers,
> >
> > Dylan
> >
> > On Tuesday 14 March 2006 12:44 pm, Ian MacMillan wrote:
> > > Dylan, I am not sure if you have this problem, but you might want to
> > > watch out with the colors when you patch landsat tiles together.  The
> > > colors tend to get screwy since the the shots were each taken at
> > > different times of day, different days of the year, etc.  I have found
> > > that it is best to enhance each tile separately, then patch them
> > > together at the end, rather than patching all band 1's, all band 2's
> > > etc.  You probably already knew this, but thought I would add my two
> > > cents.
> > >
> > > -Ian
> > >
> > > On Mar 14, 2006, at 10:04 AM, Dylan Beaudette wrote:
> > > > Markus,
> > > >
> > > > Nice images!
> > > >
> > > > a couple things:
> > > >
> > > > 1. the link at the top of the page had some interesting polar plots.
> > > > For hard copy I found that ps.rose in the GMT suite produces excellent
> > > > results as well:
> > > > http://169.237.35.250/~dylan/NPS/day100.png
> > > >
> > > > 2. the color boost technique -- is your script available on the GRASS
> > > > wiki somewhere, or in the grass scripts dir in a recent CVS branch?
> > > >
> > > > This looks like it will do the trick, now the only problem is that I
> > > > have just put together the entire state of Arizona with the poor
> > > > looking imagery!
> > > >
> > > > thanks for the tips,
> > > >
> > > > Dylan
> > > >
> > > > On 3/14/06, Markus Neteler <neteler at itc.it> wrote:
> > > >> Dylan Beaudette wrote:
> > > >>> Hi everyone,
> > > >>>
> > > >>> I was working on constructing some "true color" landsat composite
> > > >>> images today
> > > >>> from bands 3,2,1 -- and came across some strange inconsistencies.
> > > >>> Data from
> > > >>> one source produces "nice" true color images:
> > > >>>
> > > >>> http://casoilresource.lawr.ucdavis.edu/drupal/node/211
> > > >>>
> > > >>> whereas, another source produces somewhat muddy true color
> > > >>> composites. I tried
> > > >>> playing a bit with an equalized grey colormap for each band via
> > > >>> r.colors
> > > >>> color=grey.eq with the following results:
> > > >>>
> > > >>> http://casoilresource.lawr.ucdavis.edu/drupal/node/215
> > > >>>
> > > >>>
> > > >>> Does anyone have any tips on how to tweak color ramps for the
> > > >>> various bands so
> > > >>> as to get a better looking "true color" composite?
> > > >>
> > > >> how do you like this LANDSAT:
> > > >>  http://mpa.itc.it/markus/grass61/demos/
> > > >>  -> LANDSAT 7 color boost
> > > >> ?
> > > >>
> > > >> Markus
> > >
> > > -------------------------------------------------------------
> > > This message has been scanned by Postini anti-virus software.
> >
> > --
> > Dylan Beaudette
> > Soils and Biogeochemistry Graduate Group
> > University of California at Davis
> > 530.754.7341
> >
> >

-- 
Markus Neteler     <neteler itc it>       http://mpa.itc.it
ITC-irst -  Centro per la Ricerca Scientifica e Tecnologica
MPBA - Predictive Models for Biol. & Environ. Data Analysis
Via Sommarive, 18        -       38050 Povo (Trento), Italy