Novel approach for shape-based similarity search enabled by 3D PDF

Frank Neumann, Michel Atten

Abstract: During product development, shape-based similarity search helps the engineering teams to discover parts and assemblies already existing in their company's component databases that match the ones required for the new product. Therefore, similarity search constitutes an important means to keep the overall number of components designed, managed and maintained by a company at the lowest possible level. This helps reducing the efforts for engineering design, manufacturing planning, managing the master data, and consequently lowers the product costs. In this paper, we present a new approach for shape-based similarity search utilising 3D PDF as the unique neutral format for determining various types of shape descriptors. We find that this method significantly reduces the implementation efforts for analysing 3D geometries originating from various CAD systems. In addition, it permits to tackle novel use cases for shape-based similarity search due to the versatile characteristics of the 3D PDF format.