3D surface segmentation using active sensing

Manufacturing processes involving object inspection or manipulation often require 3D shape acquisition and modeling. Active sensing offers a practical solution to such problems. The authors present a new solution to extract 3D shapes from structured light images. The 2D striped image is segmented into regions corresponding to quadratic surfaces. For that purpose, a fuzzy analysis of the stripe properties is used to locate discontinuities and to characterise stripe parts. These parts are tracked in consecutive images and matched in order to create regions. The authors present segmentation results obtained with real scenes consisting of multiple objects of arbitrary shapes.<>