Research >> k-Pixel interactions
Graph cuts have proven to be
extremely useful in
vision, as they allow one to find
the global optimum,
strong local optimum, for a variety of
problems. We ask the
question: for what class of energy functions
can the graph
cuts methodology be applied, when k pixels can
interact at once?
The cases of k = 2 and 3 have already been treated
literature; we look at the general case. Many pixel interactions
are relevant in a variety of applications, including Markov Random Fields
and shape modeling.
Faculty: Daniel Freedman
Student: Matt Turek
D. Freedman and P. Drineas. Energy minimization via graph cuts: settling
what is possible. In IEEE Computer Society Conference on Computer Vision
and Pattern Recognition (CVPR), volume 2, pages 939-946, 2005.
[ pdf ] [ bib ]