[Lowfer] Post Capture Processing (was: XSR in CT)

Thu, 24 Oct 2002 20:50:09 -0400

I, too, have been dabbling in this area. But coming from a different aspect.

I have been playing with Spectrum Lab and it's ability to "dump" the FFT
buffers to a file. I am working on (simple) software that would massage the
data at a point between the FFT engine and the graphic display. This also
opens the possibility of using other plotting programs/software to display
the data.

This started as a program to compare the energy at frequencies around the
desired signal with the energy AT the desired frequency. With appropriate
averaging (still TBD) I feel that I can have the software alert me that it
is possible a signal is present. Combine this with computer control of my
Icom receivers and it is *possible* (in my twisted mind) to "scan" all
lowfers and get informed when energy (a signal) is present on the frequency
that is not present on adjacent frequencies, i.e. not wideband noise.

The software is real and I have some interesting results but nothing to
publish yet.

Thoughts?

Mike

-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of John Davis
Sent: Thursday, October 24, 2002 7:36 PM
To: [email protected]
Subject: [Lowfer] Post Capture Processing (was: XSR in CT)

>PCP = post capture processing (no drug usage here!). While not
>spectacular by any means , it may point up an area for some further
>study.
>
>Jay Rusgrove, W1VD
>

Jay,

Post capture processing does indeed show potential.  I've used gamma
correction (available even in some free graphics programs), coupled with
brightness, contrast, and sometimes even color adjustments, to manually
enhance ARGO captures for printing, and also sometimes to make the originals
"readable" where it was otherwise hard to distinguish between noise and
signal.

I first tried it a couple of years ago on some early European ham QSO
captures where it was hard to be sure what was hiding there in the
background.  By using gamma correction tools to "stretch" the contrast in
only the most important part of the transfer curve (determined on a
case-by-case basis), I found it was often possible to make a questionable
signal quite readable.

A hint to anyone trying this: start with at least 24-bit color images to
ensure good resolution between differing values.  Sometimes the subtle
changes are what count the most, so you want to preserve maximum color depth
until you have the signal well isolated.  Then you can reduce the color
depth or gray scale values for convenience.

As Jay says, it is an area for further study.

John

_______________________________________________
From the Lowfer mailing list
Send messages to: [email protected]
To sub/unsub visit: http://mailman.qth.net/mailman/listinfo/lowfer