18 May 2020, 16:00 - 17:00
We review camera architecture in the age of artificial intelligence. After a brief review of recent developments in deep learning and deep learning tools, we focus on the use of AI to control image capture (focus, exposure, etc.), to manage data capture and to fuse multisensor data. The joint impact of these technologies is to radically increase pixel capacity to gigapixel scale while also improving depth of field, dynamic range and spectral capacity. We also consider the impact of AI on physical camera design (lenses, color and polarization filters, temporal sampling).
David J. Brady is the Fitzpatrick Professor of Photonics at Duke University. He and his group built the world’s first terrestrial gigapixel camera in 2012, since that time he has continued to work through his role at Duke and his role as Chief Scientist at Aqueti, Inc. to improve camera information capacity. He is the author of the text “Optical Imaging and Spectroscopy,” and is a fellow of OSA, SPIE and IEEE. He won the 2013 SPIE Denis Gabor Award. His Ph.D. work with Demeti Psaltis focused on optical implementation of artificial neural networks, his recent work uses neural networks and compressive sensing to improve image quality while reducing size, weight and power per pixel.