Re: [CAD_CAM_EDM_DRO] 3D Laser scanners
Posted by
Jon Elson
on 2001-01-19 23:18:39 UTC
"Heuver, Brad (B.R.)" wrote:
the spinning polygon mirror in a laser printer, and replace the small IR laser
with a possibly more powerful visible laser, mostly so that you can see
the beam as you work on the thing. Then, you get some of the line-scan
CCD sensors from page scanners, and place them at strategic locations,
with objective lenses suited for the volume to be scanned. Probably
two would work for simple objects, and 4-6 would be needed for a
full 360 degree view. You mount the beam scanner and cameras on
a carriage to traverse the object in one axis. Now the tricky part comes.
You have to capture the scan lines from the cameras very fast, maybe
10 to 100 thousand scans a second from as many cameras as there are,
and have an FPGA determine the pixel where the beam hit the object.
Then, you record those intersection points. After the scan, the intersection
points are corrected for the characteristics of the optics, and the
equations are solved to determine the 3D location of each point.
This actually might not be any harder than building a gantry CNC mill,
for someone good with electronics and optics. If you want to digitize
a car body, or a transmission case, to .001" absolute accuracy, it
could be REAL tough. If you want to digitize somebody's face, or
objet d'art, and get it close enough to machine a replica, I think it's
doable.
Jon
> I went to the Detroit Auto Show this week, and at the Ford display,I can think of one way to do this pretty well. First, you take something like
> they had a 3D "human scanner". It was a booth that had a 360 degree laser
> scanner that traveled from your head to toe, and in about 5 seconds, put out
> a 3-D picture of you on a computer screen outside the booth. It captured
> loose hanging hair, folds in clothing, etc... It was a visible red light
> laser, (unless that was for show) and it took about 5 seconds to travel from
> head to toe. I looked behind the scenes, and it appeared to be all
> controlled by 1 or 2 large tower PC's. I have no idea how it worked.
> A very nice toy.
the spinning polygon mirror in a laser printer, and replace the small IR laser
with a possibly more powerful visible laser, mostly so that you can see
the beam as you work on the thing. Then, you get some of the line-scan
CCD sensors from page scanners, and place them at strategic locations,
with objective lenses suited for the volume to be scanned. Probably
two would work for simple objects, and 4-6 would be needed for a
full 360 degree view. You mount the beam scanner and cameras on
a carriage to traverse the object in one axis. Now the tricky part comes.
You have to capture the scan lines from the cameras very fast, maybe
10 to 100 thousand scans a second from as many cameras as there are,
and have an FPGA determine the pixel where the beam hit the object.
Then, you record those intersection points. After the scan, the intersection
points are corrected for the characteristics of the optics, and the
equations are solved to determine the 3D location of each point.
This actually might not be any harder than building a gantry CNC mill,
for someone good with electronics and optics. If you want to digitize
a car body, or a transmission case, to .001" absolute accuracy, it
could be REAL tough. If you want to digitize somebody's face, or
objet d'art, and get it close enough to machine a replica, I think it's
doable.
Jon
Discussion Thread
Heuver, Brad (B.R.)
2001-01-19 12:59:09 UTC
3D Laser scanners
Jon Elson
2001-01-19 23:18:39 UTC
Re: [CAD_CAM_EDM_DRO] 3D Laser scanners