Feature Based Shape Similarity Measurement for Retrieval of Mechanical Parts

Abstract

Exploiting shape similarities amongst parts for applications such as variant process planning is well known in the manufacturing industry. This particular application requires a mechanism for retrieval of similar parts from a part database which in turn requires a method for shape similarity measurement. In this paper, such a method is presented. First, the part is decomposed into simpler shapes resembling machining features. The decomposition method makes use of primitives to generate the shapes directly unlike previous methods in which the shapes are produced by combining minimal cells. Next, part characteristics that capture the spatial and dimensional relationships amongst features are used to measure the similarity. These characteristics are relevant to machining and they complement the characteristics such as feature type and feature intersections that are used by the previous shape comparison techniques. Implementation and examples are also included.