Florian Schroff, Antonio Criminisi, and Andrew Zisserman 2008
In Proc. British Machine Vision Conference (BMVC)