CSE Distinguished Seminar | David Brady

Abstract

The arbitrary segmentation of photographic systems into optics that form an irradiance image, focal planes that sample this image, image signal processing (ISP) chips that condition and compress this image and remote systems that exploit the image is demonstrably inefficient. The data cube of information passing through the aperture of a camera may exceed one petavoxel/second. The function of a modern camera is not to “image” this data, rather the camera’s function is to convert this massive but extremely compressible data stream into digital information. Detailed analysis and design of this process is especially important as the primary function of cameras shifts from the creation of images for human consumption and towards service as the eyes for AI. Over the past 30 years, the capacities of communications, computing and storage systems have each increased by factors of more than 10^6. But cameras have stayed in the megapixel range. This talk reviews novel approaches to lens, focal plane and ISP design that enable cameras to match the improvements in other information systems. In particular, we describe parallel processing in sensor arrays to increase modal capacity, feature-based space-time sampling to improve temporal resolution and dynamic range, interferometric focal planes to overcome the diffraction limit and energy efficient ISP to enable full data cube sampling.

Bio

David Brady is the J. W. and H.M. Goodman Chair of Optical Sciences at the University of Arizona, where he directs the Camera Lab in the Wyant College of Optical Sciences. Brady is a graduate of Macalester College and Caltech, where he earned a Ph. D. on the topic of volume holographic implementations of artificial neural networks. He was previously on the faculty of the University of Illinois, Duke University and Duke Kunshan University. He is the author of “Computational Optical Imaging,” which will be released by SPIE Press in the Fall of 2025 and the associated lecture series youtube.com/@arizonacamera At Duke he was the principal investigator for the DARPA AWARE program, which constructed the first terrestial gigapixel cameras. He also received the SPIE Denis Gabor Award and the Optical Emmett Leith Medal for the development of compressive holography.