(Georgia Institute of Technology, 2008-01-24)
Barbieri, Nicholas P.
Lenticular sheets can be used with machine vision to determine
relative position between two objects. If a lenticular sheet of a
given period is mounted above periodically spaced lines sharing the
same period, lines will appear on the lenticular sheet which
translate along the lenticular sheet in a direction perpendicular to
observer motion. This behavior is modeled theoretically and tested
experimentally, and found to be linear within a finite range.
By arranging two lenticular sheets, configured as described above,
in a mutually orthogonal configuration on a flat surface, the lines
that appear on the lenticular sheets can be used by a camera to
estimate its position relative to the lenticular sheets. Two such
devices were constructed to test the principle, and machine vision
code was developed to ascertain position using these devices.
Machine vision code was found to reliably provide angular position
of a camera within $1.4^circ$ through experimental testing.
The optical patterns that appear on the lenticular sheet surfaces
are monitored using a digital camera. The resulting images are
analyzed using visual C++ in conjunction with the OpenCV library and
the appropriate camera device drivers. The system is able to
estimate height, yaw, and position relative to the optical target in
real time and without the need for a prior reference.