ATM Computerized Foucault Test

[Hennecke Marcus <Marcus.Hennecke@temic.com>]

With some people attaching video cameras to their Foucault tester, I had an
idea: Basically, in the classic Foucault test, one has to compare the grey
level of two areas opposite of one another. This is usually done through a
Couder mask. Comparing them and making sure that they are the same is not
always easy. If the video camera were connected to a computer, could one not
let the computer make the decision which zone we are currently looking at?
This web cameras are getting cheaper and cheaper and should probably suffice
for this.

Moreover, one could leave the mask away and do the following: For increasing
radii starting at the center of the mirror, compute the variance of the grey
level along the perimeter of that radius (i.e., the variance within a zone
of that radius). The knife edge should then be at the ROC of the radius with
the lowest variance.

Now add an encoder to the micrometer screw of the tester. The computer can
then also measure the position of the knife edge.

The testing procedure could then look as follows: Move the knife edge until
it is at the ROC of the center zone. Let the computer do a reading of the
position. Then move the knife edge a bit and let the computer do a reading
of the position and finding the corresponding zone. Keep doing this until
you reach the edge zone. With the right math it doesn't matter too much if
you measure the same zones each time since the computer knows which zones
they are.

Would this work?

Of course to make it work there are some details that need to be solved: The
computer needs to find the center of the mirror in the image. Finding the
radius with the lowest variance might not be as exact as it sounds in
theory, there might be several local minima, the minimum might be very
broad, etc. Some smoothing might be necessary.

Comments anyone?

Marcus

--
Dr. Marcus E. Hennecke
TEMIC Telefunken Microelectronic GmbH
Speech Processing L/B2/S/EA-U  Tel +49-731-3994-117
Soeflinger Str. 100            Fax +49-731-3994-61-117
89077 Ulm - Germany            marcus.hennecke@temic.com
Private: http://www-isl.stanford.edu/~marcush/