Extracting Freeform Shape Information by Template Fitting

Finding effective and interactive tools for extracting freeform shape information continues to be a challenging problem in reverse engineering. Given a freeform shape, it may be constructed by adding one shape, named a pattern, to another. In this paper, an approach of extracting the pattern by template fitting is proposed. By similarity analysis, a user-defined region of interest in the shape can be matched, or fitted, to a shape template. According to the different methods in constructing the shape, several different kinds of R³ to R³ functions are defined. With those functions, the original shape is mapped to the fitted shape template, thus the template can be used as a “ruler” to measure the region of interest in the shape. The measuring results, e.g., the extracted pattern, can be generated through an inverse mapping, thus it can be used in the future design. Several implementations were conducted based on ACIS^® and OpenGL^® in order to verify the proposed method. It is also described how the proposed technique can be applied in practical shape modeling applications.