Geometric Sensors on Optical Fields

David J. Brady, Duke University

Sensing on radiation fields involves a mapping between sources distributed in continuous spaces and discrete measurements. Conventionally, the measurements have been represented as samples on a space isomorphic to the source space or to a continuous transformation of the source space. By "geometric sensors," we mean systems in which the vector space of the measurements is not trivially embedded in a space isomorphic to the source space or simple transformations thereof. Recent work on artificial noses provide a particularly clear example of geometric sensors.

Potential advantages of the use of geometric sensors on optical fields include direct embedding of data and image compression in the physical layer, efficient capture of spatial and spectral features and multidimensional object tracking. The Duke Integrated Sensing and Processing group is developing geometric sensors for human tracking, 3D microscopy and Raman spectroscopy. Fabrication of complex three-dimensional sensor elements based on photonic crystals, reference structures or holograms are central to these efforts.

This talk describes the motivation for geometric sensing, presents two examples of geometric design (reference structure tomography and multimodal multiplex spectroscopy) and describes programs for further development.

The P/T Colloquium is
typically held each
Thursday, 3:45–5:00 PM.
Refreshments are served
at 3:15 PM.