| Re: Warping with intensity conservation or drizzle algorithm [message #60827] |
Fri, 20 June 2008 14:12 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
> (2) JD Smith's CUBISM software (http://turtle.as.arizona.edu/jdsmith/
> cubism.php ) uses an algorithm similar to drizzling to resample 1-d
> spectra. The workhorse procedure for this algorithm is called
> polyclip.pro and it performs Sutherland-Hodgman polygon clipping.
> He also provide a C version which is 50 (!) times faster. (So
> let's encourage ITTVIS to include Sutherland-Hodgman clipping or
> something similar as an intrinisc IDL function.)
Polyclip.pro and its associated polyfillaa.pro and polyclip.c could
readily be re-worked for a drizzle-like algorithm (by adding its
somewhat poorly justified "drop size" reduction). It is coded in C,
but IDL auto-compiles it, which seems to work well on most platforms
(though it requires you to have a C compiler). It also has a much
slower version written in IDL native code to fall back on if it can't
compile it.
I've pinged RSI/ITT about integrating a general flexible clipper, like
GPC, many times, but I don't think it's reached critical threshold.
Add your voice to the chorus. Polyclip is very simplistic and special-
purpose, and for real speed, you need to stuff a huge number of
polygons down its throat during a single call to CALL_EXTERNAL. For
added perversity, it uses multiple REVERSE_INDICES style inputs and
returns for the resulting clipped polygon indices. That said, it does
fly through millions of polygons in a very short amount of time.
JD
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
