Re: gEDA-user: SMT assembly and gerbv lilbro.mosquito.net.nz

[Dave N6NZ <n6nz@xxxxxxxx>]

John Griessen wrote:
> Dave N6NZ wrote:
> maybe simply
>> looking at the gerber aperture and stroke information and grouping pads 
>> by identical shape and identical spacing is a good first footprint 
>> grouping.
> 
> The silk needs to be considered to get useful groups.
> I'm no AI coder, and don't plan on learning to do that
> for this task.
> 
> Even rasterizing footprint at a time to do a fuzzy comparisons seems like a lot of
> coding for the value...

You might take a look at the Intel OpenCV library:

http://www.intel.com/technology/computing/opencv/

This is very mature code, with lots of active development. A few years 
back I worked at Intel in the same division, and the lead developer was 
a few cubes down.  One day he got an e-mail from a guy in the fab that 
had downloaded it and built a wafer sorting application on this library. 
  It might be possible to use this library for the footprint recognition 
problem, by training it on a library of footprint examples or something. 
  I'm not to sure -- I never dug into the details of the API.  At the 
time, I was a PHB, so didn't get to play.  But the code is widely used.

I glanced at the license, and it appears to be BSD'ish.  Intel uses 
different open source licenses to achieve different objectives, so you 
have to look project-by-project. (The division attorney was 2 cubes 
down... I managed the product validation group so all releases from the 
division came through me... the attorney and I had a lot of interaction.)

Another quick glance showed that OpenCV may be built on the IPP (Intel 
Performance Primitives) library, which doesn't appear at first glance to 
have an open license.  But the IPP is just an Intel-tuned BLAS back-end 
-- hand tuned vector primitives and such -- so it can always be replaced 
with a generic alternative since all layers of the BLAS API's are public 
standards.  I believe open source generics exists for the the entire 
stack.  (The IPP guys were in Oregon, so I only got to hassle them at 
release time, for the most part. :)

-dave

> 
> John G
> 


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user