Data segmentation and model selection for computer vision. A statistical approach (Q4935252)

This book is a collection of articles about segmentation and fitting and it is divided in three parts. Apart from the first chapter (part one) that gives an historical survey of the segmentation problem, the contents are focused on robust fitting and model selection. Part two is about statistical foundations of the methods and is of very general interest. The three chapters mainly treat of the model selection problem (how can we ensure that variables in a (perhaps too) big model are (not) relevant) and the robust fitting (possibility to treat datas that present deviations from the assumed model). The two chapters consisting part three deal with segmentation and model selection. The first one investigate methods based on \(K\)th order statistics but avoiding to make any optimization to find the right \(K\) value. The last chapter explores ways of automating model selection when the problem is structure and motion recovery from multiple images.