抄録
Large curved steel plates are often used for ships and storage tanks. They are gradually manufactured from flat plates in many bending steps. If their curvature and deformation can be measured in each step, the quality and process time will be improved. Laser scanners can be used to capture dense points on a plate and to calculate geometric properties of the plate. In this paper, we consider supporting plate-bending processes using dense point-clouds. To support bending processes, it is necessary to identify specific positions on plates while the workers are bending a plate. We solve this problem by generating a mesh model of a plate using a point-cloud and parameterizing it on the normalized space. In our experiments, the properties of plates could be very quickly calculated from point-clouds with practical accuracy.
