Re: Accurate edge detection?

On Jul 26, 9:20 pm, slus...@[EMAIL PROTECTED]
 wrote:
> On Jul 25, 2:15 pm,illywhacker<illywac...@[EMAIL PROTECTED]
> wrote:
> > On Jul 23, 11:12 pm, slus...@[EMAIL PROTECTED]
 wrote:
> > > On Jul 23, 12:01 pm,illywhacker<illywac...@[EMAIL PROTECTED]
> wrote:
> > > > On Jul 17, 7:54 pm, slus...@[EMAIL PROTECTED]
 wrote:
>
> You said: "If you know all the relevant
> parameters, your life will be much easier, otherwise you will have to
> infer these too."
>
> Indeed!  Thanks very much for your time and advice.

Dear Spencer,

I am afraid that the problem you seem to think is relatively trivial
is far
from being so. You have the air of someone who thinks they know more
than
the people they are talking to, but alas a little knowledge is a
dangerous
thing.

> You said: "If you know all the relevant
> parameters, your life will be much easier, otherwise you will have to
> infer these too."
>
> Indeed!  Thanks very much for your time and advice.

You seem to find this flippant or amusing, but it is the heart of the
problem. How much information do you think you will need to predict
the
image using physics? First of all, you need to define 'edge' more
precisely. Then you need to know the geometry of the particular edge
you
are dealing with, which may not boil down to a mere 'position', you
need to
know the illumination, its spectral content, etc., and you need to
know the
properties of the camera. How much of this information are you
assuming you
know? Are you assuming a simplified toy example, or some conditions
related
to a particular type of experiment? Which approximations are you going
to
make? Not geometric optics, obviously, since you speak of diffraction.
Which of these parameters do you imagine will not affect the image
formed
by the edge?

If some of the parameters are unknown, then you will have to deal with
them
in addition to the edge 'position'. There are well-defined
mathematical
ways of attacking this problem with probability theory, apparently
unknown
to you. If you can get over the idea that you are talking to a bunch
of
hick image processors, then I will say more.

illywhacker;