Re: Optical - servo alignment or "registration"

In article <baac3033-26fa-44fc-8dc1-3035d1dd1ba8@xxxxxxxxxxxxxxxxxxxxxxxxxxx>, cs_posting@xxxxxxxxxxx wrote:
On Mar 21, 1:09 am, Spehro Pefhany <speffS...@xxxxxxxxxxxxxxxxxxxxxxx>
wrote:
For a number of purposes (eg. registration of printing) it's useful to
be able to servo a X-Y-theta platform to center multiple markings of
some kind in an couple of arbitrary areas viewed by a digital camera
or cameras-- in the printing case, we'd want to align a substrate
(perhaps with previous impressions already having been made) with the
printing screen, stencil or plates.

Growing up in the digital age, I always assumed that real world
systems did this kind of thing
by connecting a 2d camera to a computer, and then "thinking" about the
picture.

Then I learned that so many inherit from older ideas - often 1d
sensors, setups designed to give basically binary contrast (or peak
searches to do that), triangular masks over photocells, position
sensitive devices, etc...

What I'm trying to say is that there's the fancy "imaging processing"
solution, and then there are the simple ones that find a single mark
and calculate a deviation and go from there. Similarly, there are
software solutions to the noise problem and there are use a really
bright light and lots of flat non-reflective shields type ones.

Probably comes down to which is more fun / cost effective - building
physical apparatus or writing software.

There was a project yet to be completed, using a imaging system
which also included depth. It did focus on points. Software was written for
this. The part not completed was the 6 axis servo system moving a persons body
to keep a PET scan in line.......

greg
.

Quantcast