Building a smooth and well structured surface to fit unstructured 3-D data is always an interesting topic in Computer-Aided Design (CAD). In this paper, a method of approximating complex freeform shapes with parameterized freeform feature templates is proposed. To achieve this, a portion of a digitized 3-Dimensional (3-D) shape should be matched, or fitted, to a deformable shape feature template, where the deformation is a function of intrinsic feature parameters. 3-D shape matching to, possibly sparse, inaccurate or otherwise degraded, freeform surface data is known to be hard. Using a variant of the directed Hausdorff distance measure of shapes, it is shown that convergence towards a shape match is feasible. Based on sensitivity analyses of the shape distance measures, it is determined that adjusting coefficients of the optimization function in different stages of optimizations can accelerate the optimization procedure. By the matching results, a standard deviation-like function is proposed to achieve automatic feature recognition. With the proposed extendable concept, complex freeform shapes are tracked and fitted automatically. Based on a defined interference ratio, interfered feature can also be identified. Numerical experiments were conducted in order to verify the proposed method and to find the maximal degree of feature interference for which matching is successful. It is also described how the presented technique can be applied in shape modeling applications.

This content is only available via PDF.
You do not currently have access to this content.