On Nov 19, 2011, at 11:55 PM, chm wrote:

> On 11/19/2011 10:55 PM, Puneet Kishor wrote:
>> ..
>> 
>> Also of interest -- I recently stumbled upon an image IO library
>> called Bio-Formats (developed right here on my campus). It is
>> Java-based, so I don't know if it can be embraced by PDL, but it can
>> read many, many different scientific image formats, and write out a
>> few standard ones. More info on Bio-Formats is at
>> http://loci.wisc.edu/software/bio-formats
> 
> Maybe the Java or Java::Import module would be of
> use to you here.  I haven't used these modules but
> the documentation look good


The Java::Import module seemed promising, but I could not install it because 
its dependency GCJ::Cni failed to install. Looking at its build reports, seems 
like it has 2 PASS vs. 340 FAIL reports 
http://www.cpantesters.org/distro/G/GCJ-Cni.html#GCJ-Cni-0.03


> for what you are wanting
> to do...
> ..


My main interest is to be able to read a small basket of raster formats that I 
routinely stumble upon (ERDAS Imagine, Arc GRID, GeoTiff, NetCDF, perhaps a 
couple other) and be able to programmatically query data from them. Usually the 
queries are trying to find values at a location (point or polygon) or construct 
some derivation (histogram and other statistics).

When I get a new dataset, ideally, I wouldn't have to transform it to something 
else in order to make it queryable, because that would add another step, more 
complexity, in the process. And, ideally I would have to deal with only one 
query tool -- PDL would serve me well because I know a little bit of Perl. If I 
have to keep on shuffling between Perl and Python and Java and something else, 
it just makes the job harder.

Here is an example of something I have done --

1. Go to http://mumbai.geology.wisc.edu/waterisotopes

The page will prompt to allow you to use your location... press OK

2. A map comes up showing isotope ratios of hydrogen and oxygen. These data are 
in ERDAS format on my computer. I am using GDAL via MapServer to convert them 
into the images.

3. Click on the map to draw a small polygon; close the polygon by 
double-clicking.

A small popup comes up showing the histograms, min, max, mean and sd of the 
isotopes in the polygon that you drew.

To accomplish the above, I use a couple of gdal utilities called from my Perl 
program (my web applications are written in Perl). The output from the gdal 
program that calculates the statistics is written out to a file that I then 
parse with Perl and extract the values, convert them to JSON and send them back 
to the browser where the charts and the popup are constructed.

Unfortunately, the gdal programs are rather general purpose, so they can only 
do rectangular queries. So, even if you drew a triangle on the map, you will 
still get the stats for the minimum bounding rect around the poly that you 
drew. Plus, I can't do any other analysis unless I knew C++ and rewrote the 
gdal programs.

Converting rasters to piddles would provide a great amount of flexibility, but 
PDL is currently limited in the different formats it can ingest.

--
Puneet Kishor
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to