Our understanding of object recognition can address the needs of only the m
ost stylised applications. There is no prospect of the automated motorcars
of Dickmanns et al. knowing what is in front of them anytime soon; searcher
s for pictures of the pope kissing a baby must search on a combination of t
ext, guesswork and patience; current vision based HCl research relies on hi
ghly structured backgrounds; and we may safely guess that the intelligence
community is unlikely to be able to dispense with image analysts anytime so
on. This volume contains a series of contributions that attack important pr
oblems in recognition.