TECH: Optical Tracking in Human-Computer-Interfaces

Franz Madritsch

unread,

May 19, 1994, 7:35:09 PM5/19/94

to

Optical Tracking in Human-Computer-Interfaces

Progress in computing power in combination with live video
input is opening new possibilities for optical tracking.

Using 2 SGI Indys I am doing visual tracking of LED-Markers,
utilizing SGI standard IndyCams. In first experiments with
this new input medium we are already able to support a number of
applications such as:

- Manual digitizing of surfaces (e.g. face) by a laser-pointer
- 3D LED-pointer as input device
- Tracking of possition & orientation by tracking multiple LEDs
at about 20 Frames/s

Future effort will be made in animation of digitized faces including
texture and synchronisation to a speaker.

Please let me know if somebody is making progress in optical tracking
or knows about related publications.

-----------------------------------------------------------------------
Franz Madritsch
Graz University of Technology
Institut for Computer Graphics Tel.: +43 +316 841766
Muenzgrabenstr. 11 Fax.: +43 +316 846304
A-8010 Graz
AUSTRIA email: fr...@icg.tu-graz.ac.at
-----------------------------------------------------------------------

Dave Stampe

unread,

May 23, 1994, 1:19:23 PM5/23/94

to

fr...@icg.tu-graz.ac.at (Franz Madritsch) writes:
>Please let me know if somebody is making progress in optical tracking
>or knows about related publications.
>

Dan DeMenthon at University of Massachusetts is working on a peripherial
he call the light mouse: 4 LEDs on a pen, tracked by a camera.

I work with optical tracking using special hardware and IBM PC's to
do the image processing. I can track up to 100 points in real time,
computing target XYZ and pose with special algorithms. I also
work with pupil tracking for gaze monitoring systems, using the
same hardware, and track LEDs with the same software to correct
gaze position for head motion.

Lately, I am finishing up a project which tracks LED and eye position
at 250 samples/second. Had to build the cameras from scratch, but it's
well worth it.

Daniel DeMenthon

unread,

May 25, 1994, 2:26:01 PM5/25/94

to

Dave Stampe (dst...@psych.toronto.edu) writes:
>Dan DeMenthon at University of Massachusetts is working on a peripheral
>he calls the light mouse: 4 LEDs on a pen, tracked by a camera.

I (Dan DeMenthon) am actually with University of Maryland
(dan...@cfar.umd.edu). We have been calling our system Video Mouse, but I
see the name is already taken by Edmund Scientific for quite a different
gizmo (see back cover of their Optics 1994 catalogue). So OK for Light
Mouse, Dave :-) We built our first prototype in 1992 using a Steve Ciarcia
ImageWise Transmitter (now *discontinued*) that we modified (ROM and
hardware) to output centroids of bright spots for each video field
(60/sec). We now have a grant to develop an inexpensive dedicated camera
with Data Design, Inc. of Gaithersburg. The new design does not
produce an NTSC signal and will output centroids of bright spots to a
serial line (180 fields per second). We'll have something to show in
two months.

A couple of weeks ago, I posted a message enquiring for
sources of board-mounted lens assemblies for this project. None of the
companies that people suggested by email actually sold the whole
assembly. However one company sells a part of it --the lenses in a 12mm
dia. barrel with a 1/2 mm thread. They are now trying to track down a
manufacturer for the other part --the little box that encloses the CCD
chip and has holes for screwing directly to the PCB, and a threaded 12 mm
hole on top where the lens barrel is mounted like a miniature C-mount
lens. I'll post their results in a while.

Daniel