Phs. Torr, GEOMETRIC MOTION SEGMENTATION AND MODEL SELECTION, Philosophical transactions-Royal Society of London. Physical sciences and engineering, 356(1740), 1998, pp. 1321-1338
Motion segmentation involves clustering features together that belong
to independently moving objects. The image features on each of these o
bjects conform to one of several putative motion models, but the numbe
r and type of motion is unknown a priori. In order to cluster these fe
atures, the problems of model selection, robust estimation and cluster
ing must all be addressed simultaneously. Within this paper I place th
e three problems into a common statistical framework; investigating th
e use of information criteria and robust mixture models as a principle
d way for motion segmentation of images. The final result is a general
fully automatic algorithm for clustering that works in the presence o
f noise and outliers.