Research >> Illumination-invariant tracking
Tracking through sharp illumination changes is a challenging task, particularly
if the algorithm relies on photometric information. We pose the problem based
on computing optical flow in an illumination-insensitive manner, using the "pairs
of pairs" technique. The relevant energy function can be minimized using graph cuts,
yielding a highly robust algorithm.
Faculty: Daniel Freedman
Student: Matt Turek
D. Freedman and M. Turek. Illumination-invariant tracking via graph cuts. In IEEE
Computer Society Conference on Computer Vision and Pattern Recognition (CVPR),
volume 2, pages 10-17, 2005.
[ pdf ] [ bib ]