ABSTRACT
The paper presents an automatic, coarse method for 3D modelling of metal objects with complex geometry for a need of volume estimation. The field research were conducted on a historic wrought iron bar that covers the historic well in Nysa (city In southern Poland). The presented modelling methodology is based on a layered convex-hull method, which involves dividing of a point cloud on the segments. Within each segment, segmentation is performed based on the minimum distance between points. The resulting sets of points are then modelled as a convex solids. Thanks to the segmentation of point clouds in each segment and the integration of convex shells a detailed object model can be obtained. That allows to estimate the geometric parameters such as volume and surface area of the object. The advantage of the proposed method is that it has a small number of parameters: a thickness of segment and the parameter of maximum distance between points in the process of segmentation of clouds within the segment. Applying the convex hull algorithm causes a selective filtering point clouds, thus resulting 3D model is based on a much smaller number of vertexes than the initial number of points in the cloud. The disadvantage of the proposed algorithm is an irregular triangle mesh models, resulting in low surface regularity and larger items, and sensitivity to measurement errors (noise, ghost points).