kwkarth:
>Hey! Where's my "atta boy?" (Blr deserves one too) I was
>the one who grasped your need and suggested a specific and
>affordable solution to your quandry in the first place!
You're right, my apologies. I forgot the "Always bring
enough for the whole class" rule.
Thank you kwkarth and
blr, for not being the cause of the great length of this
initially simple thread.
ponzio:
>Here are two papers which may interest you.
Thanks for the links. These papers should prove invaluable
in clarifying my point for you. Here's a quote from the
first one:
"Perfect denoising is not possible: the higher the
threshold coefficient is set, the more noise is detected,
but the more of the original signal is affected as well."
As should be clear, what is outlined in the paper mangles
the signal, especially if it naturally contains areas of
white noise. If the idea of allowing the signal to be
mangled just to remove the noise is taken to the ultimate
extreme, you'll find that one can completely eliminate the
noise by simply turning off the TV.
Meanwhile, my $5 solution does absolutely nothing to mangle
the signal, yet still reduces the hiss to inaudible levels.
Here's a quote from the second one:
"The only assumption we make is that the signal was
generated by a dynamical system with a small number of
degrees of freedom."
See this? This is a perfect example of an assumption about
the nature of the signal.
>I appreciate the difficultly you have grasping the scope
>of your problem.
I have no diffculty "grasping the scope" of my problem. I
understand it fully, as do blr, kwkarth, and pedxing. In
fact, as far as can be seen from this thread, you are the
lone hold-out, in your overestimation of both the problem
and its solution.
>While it is true that you cannot determine a random number
>when combined with another random number. That is not the
>case in your situation and in many others. It is the
>"non-randomness" of the noise and the data signal which
>allows noise removal.
Actually, the noise IS random, and the signal may at times
be random as well, so it IS my case, at least part of the
time. This is another perfect example of one of those
assumptions about the nature of the signal that I was
talking about.
>A quick search of the US Patent database using the terms
>(remove or reduce and "white noise") yielded 1221 patents
>granted in the last six years. While that is too many to
>examine one by one, 6,332,136 "Fuzzy filtering method and
>associated fuzzy filter" is on point and is the most
>recently granted.
So now we're going to decide what is and is not possible
based on the judgement of the patent office? I would point
out the MANY patents granted to impossible perpetual motion
devices, but here we are talking about impossible
algorithmic patents, so I'll point out a patent more
germane to this discussion. Take a look at this analysis
of patent 5,533,051:
http://www.teaser.fr/~jlgailly/05533051.html
>Tempus, it has been a pleasure corresponding with you.
It has been an interesting experience for me as well.
>I wish you luck in your future technological education.
Yes, my education of the uninformed masses is often an
uphill battle. Sometimes they refuse to listen.
