The wide availability of scanning technologies for 3-d shape acquisition has led to the development of algorithms and software, in both academia and industry, for building surface meshes from the raw data returned by the scanner. For most optical and contact sensors this raw data is in the form of an unorganized point cloud, or PCD set (point cloud data); current scanners are able to generate object surface samplings at submillimeter accuracy. At such high resolutions, these dense PCD sets become adequate representations of object geometry in themselves, without any need to fit a mesh or surface. The work undertaken under this grant will exploit tools from algebraic topology and computational geometry to perform local and global analysis directly and efficiently on PCD data. Specifically, algorithms for PCD feature detection, segmentation, matching, fitting, and compression will be developed and analyzed.
It is expected that PCD sets will become a lightweight and convenient representation of object geometry not only in reconstruction or visualization applications, but even more so in the largely unexplored area of directly interrogating object geometry in the PCD format for applications such as robotic exploration or product customization. Over the next five years we can expect to see wide commercial deployment of 3-d scanning and the appearance of compact, portable scanning devices. By exploiting the acquired PCD data it will become possible to customize and tailor manufactured products to their user to a degree that has been impossible up to now. Technology developed under this proposal should facilitate this transition in the apparel, footwear, hearing-aid, orthodontic, prosthetics, and automotive industries.
