🚀 go-pugleaf

In article <gotwols.713750652@warper.jhuapl.edu> gotwols@warper.jhuapl.edu (Bruce Gotwols) writes:
>I have used both IDl and Matlab for both signal processing and 
>image processing.  As others have correctly said, Matlab is stronger
>for signal processing and IDL for image processing.  However, IDL has
>one important feature that caused me to settle on it for almost all of
>my work in both areas:  It can handle bytes, 16  and 32 bit integers, 
>floats, doubles, and complex.  Matlab is restricted to only doubles and
>complex....a very severe restriction if you must work with huge arrays
>as I often do.

What you say is true for MatLab versions < 3.5i, but MatLab 4.0 is supposed
to handle all the data formats you mention.

>IDL allows routines written in other languages such as C and Fortran to
>be hooked into it.  Therefore, I write for instance, digital filter programs
>in C and then hook them into IDL at run time.

This can be done with MatLab 3.5i too.  There are demo routines provided
(e.g., engine.c) that allow MatLab to communicate with other routines via
UNIX pipes and/or files.

>  I also have written a set of
>routines that allow me to manipulate Exabyte tape drives from within IDL (using
>the hooks mentioned above).  This way I can analyze several Gbytes of radar 
>signal data straight from tape without having to first put the data on disk.
>I do not know if this can be done from Matlab; perhaps someone will comment 
>on this.
>
I haven't tried controlling our Exabyte drives from within MatLab, but I don't
see any difficulty in doing so.  Using the MatLab "engine" and UNIX shell
escapes (!), it should be doable.

>There is an anonymous ftp site with hundreds of well written IDL routines
>of general interest.  It is fermi.jhuapl.edu.  Get into idl-pvwave and then
>the next level down which I think is called jhuapl.  By the way, most of these
>routines will also run under pv-wave, but I prefer IDL.
>
Of course there are also vast amounts of MatLab code available (e.g., netlib).

There are many things I like about MatLab 3.5i (I'm waiting for 4.0).  However,
it doesn't do everything.  I'm surprised that a couple of the other systems I
use haven't been mentioned yet in this thread.  For 1-D signal display, I often
prefer Entropics' ESPS/waves+.  For symbolic signal processing, I use
Mathematica 2.1 with Brian Evans' Signal Processing Packages 2.71.  Although
MatLab, Mathematica, and ESPS/waves+ are expensive, hopefully you could
convince people, how could you not afford to have them?

Joe
-- 
.............................................................................
; Joe Campbell  N3JBC  jpcampb@afterlife.ncsc.mil  74040.305@compuserve.com ;
; My opinions are mine!   Happiness = Reality - Expectations, Click & Clack ;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~