Quote:
Originally Posted by rambo1152
No, I rather expected to be castigated (ooh er missus), by a proper Engineer such as your good self, for adopting such an empirical method.
|
No, I think it's sound engineering practice! Well, there's a bit of psychology in there too -- the cunning part is making sure the psychology is working in our favour. You say "adjust for maximum noise", but what you are inevitably actually going to be doing is adjusting for maximum (signal plus noise).
Noise is just noise, with no structure to it. It just looks like noise. It will always have exactly as much light as dark in it, no matter how much gain you apply, because that is the nature of the beast: there is exactly as much energy above the line as below, and the average over time is zero. Noise
riding on an actual signal looks worse than just noise, because there are still some discernible fragments of structure in it; and these are perceived subjectively worse than just random, structureless noise. So when the noise is as bad as you can get it, that means there must be a good strong signal lurking somewhere nearby. You are basically making the target for your initial search easier to hit -- it's easier to get it right once it's already nearly right, and a broken-up picture with lots of noise on it is more nearly right than just noise.
It's like the old joke about the physicist, the engineer and the mathematician who are all staying in a hotel one night, when a fire breaks out in each of their rooms. Each of them wakes when the smoke alarms sound. The physicist does some quick calculations involving enthalpy of combustion, specific heat capacity and latent heat; then pours on exactly enough water to extinguish the flames, yet have every drop of it evaporate from the residual heat. The engineer keeps pouring on water until the fire is out; then pours on twice as much again, just to make sure. The mathematician yawns, notes that this is identical to a previously-solved problem, and leaves it as an exercise for the reader .....