OSC Colloquium: David Brady, "Neural Processing and Snapshot Compressive Imaging"

Title: Neural Processing and Snapshot Compressive Imaging

Abstract(s): 

Most imaging systems capture measurements distributed over two dimensional sensor planes. Most objects of interest are 3, 4 or 5 dimensional. (Hyperspectral images, video and volume stills are 3D, adding time, space and frequency together yields 4D or 5D data cubes). Over the past decade, snapshot compressive imaging or compressive tomography based on sparsity constraints has been applied to remove dimensionality trade-offs and enable efficient estimation of high dimensional data from 2D measurements. Recently, neural estimators and hybrid algorithms have radically improved decompressive inference. This talk reviews diverse compressive tomographic imaging systems and explains evolving sampling system and algorithm designs for such systems. 

Speaker Bio(s): 

David Brady is the J. W. and H. M. Goodman Professor of Optical Sciences in the Wyant College of Optical Sciences. He is a Fellow of OSA, SPIE and IEEE and received the 2013 SPIE Denis Gabor Award for the development of compressive holography. He earned a B.A. in physics and math from Macalester College and M.S. and Ph. D. degrees in Applied Physics from Caltech, where his thesis was on the topic “Volume holography in artificial neural networks.” Brady was previously on the faculty of the University of Illinois and Duke University. He enjoys running, crossword puzzles and, obviously, photography.