This Small Grant for Exploratory Research provides funding to explore the feasibility of physics-based similarity detection algorithms for CAD models. A CAD model is an expensive and important part of the whole production process and there exists a natural need to protect the ownership of this data model. An equally important but closely related issue is the efficient indexing of CAD models databases, which could have significant financial implications as well. The novel approach we propose is very similar to spectroscopy, which employs an extraordinarily indirect means by observing the pitches at which a molecule naturally vibrates (i.e. its natural frequencies or spectral fingerprints) to determine the composition of stellar bodies. If it proves possible to define, formulate, and identify such spectral fingerprints of a CAD model (in the context of Structural Dynamics) to distinguish it from other CAD models, the impact would be significant to the engineering design community. The embedding of the CAD model geometry within a vibratory model is the overall formal procedure that would be goal of the proposed study. In our SGER endeavor, we plan to focus on the planar and free-form surface models as our preliminary work.
Successful development of this project will have broad impact on our modern information society. One such promising area is copyright protection of CAD models. Another application field of efficient and shape intrinsic search of CAD model databases is useful in diverse disciplines such as mechanical engineering, bio-informatics, medical imaging, and computer vision. Our applications are of further interest to the computer graphics community and can have broader effects to society at large such as automated inspection in manufacturing, product life cycle management, Google-like database search engine for CAD models, registration and object recognition for military or homeland security purposes, and recognition of morphing shapes in weather prediction.
