3D Shape Modeling Employing Fuzzy Clustering for Stereo Vision


In this paper, a novel fuzzy-clustering-basedmethod for 3D shape modeling is proposed. This method isintended for scenes involving multiple objects, where eachobject is replaced by a primitive model. The proposed method iscomposed of three stages. In the first stage, 3D data isreconstructed using stereo matching technique from a stereoimage taking multiple objects. Next, the 3D data is divided intoa single object by employing a Fuzzy c-Means augmented withPrincipal Component Analysis (PCA) and a criterion about thenumber of clusters. Finally, the shape of each object isextracted by Fuzzy c-Varieties with Noise Clustering.

Sheraton Vancouver Wall Centre Hotel, Vancouver, BC, CanadaJuly 16-21, 2006


Details 詳細情報について

Report a problem

Back to top