Abstract
This paper extends the “cross-sectional” approach for reverse engineering, used abundantly in biomedical applications, to the mechanical domain. We propose a combination of “projective” and cross-sectional algorithms for handling physical artifacts with complex topology and geometry. In addition, the paper introduces the concept of constraint-based reverse engineering, where the constraint parameters could include one or more of the following: time, storage (memory, disk-space), network bandwidth, Quality of Service (output-resolution), and so forth.
We describe a specific reverse-engineering application which uses ultrasound (tilt-echo) imaging to reverse engineer spatial enumeration (volume) representations from cross-sectional data. The constraint here is time, and we summarize how our implementation can satisfy real-time reconstruction for distribution of the volume data on the internet. We present results that show volume representations computed from static objects. Since the algorithms are tuned to satisfy time constraints, this method is extendable to reverse engineer temporally-varying (elastic) objects. The current reverse engineering processing time is constrained by the data-acquisition (tilt-echo imaging) process, and the entire reverse engineering pipeline has been optimized to compute incremental volume representations in the order of 3 seconds on a network of four processors.