ECE Seminar with Vivek Goyal
- 11:00 am on Wednesday, June 5, 2013
- Photonics Center, 8 Saint Mary’s St., Room 339
Signal Processing for Novel Acquisition Technologies With Vivek Goyal Massachusetts Institute of Technology Abstract: Signal processing plays a central role in exciting new information acquisition technologies. I advocate an integrated view of signal modeling, physical device modeling, and algorithms for inverse problems. The sampling of results provided in this talk will exemplify improved theoretical understanding of inverse problems, improved performance in established applications, and innovative system architectures for acquisition that are inspired by this holistic view. The talk will concentrate on the following two main topics: Compressed sensing has brought the use of sparsity- and compressibility-based signal models to the forefront of data acquisition and inverse problems. Inspired by the conservatism of the well-known analyses of compressed sensing, we develop instead a Bayesian analysis framework. Under the common assumption of replica symmetry, we prove a convergence result that provides a simple scalar equivalent problem from which one can make various asymptotically-exact performance computations. The method applies to least squares estimation with any separable regularizer, including but not limited to the l1 regularization that has been extensively studied of late. The accuracy of our analysis makes it a useful system design tool, and the analysis also inspires new algorithmic approaches. LIDAR systems and time-of-flight cameras use time elapsed from transmitting a pulse and receiving a reflected response, along with scanning by the illumination source or a 2D sensor array, to acquire depth maps. We introduce a method for compressive acquisition of scene depth with high spatial and range resolution using a single, omnidirectional, time-resolved photodetector and no scanning components. In contrast to compressive photography, the information of interest – scene depths – is nonlinearly mixed in the measured data. To overcome this aspect of the inverse problem, the depth map reconstruction relies on parametric signal modeling of the optical impulse response of piecewise-planar scenes. Through the use of parametric deconvolution, we achieve much finer depth resolution than suggested by the illumination pulse width and detector bandwidth alone. This opens up possibilities for compact, low-power 3D sensing for applications such as gestural interfaces and augmented reality in mobile devices, which we discuss as well. About the Speaker: Vivek Goyal was a Member of Technical Staff in the Mathematics of Communications Research Department of Bell Laboratories and a Senior Research Engineer for Digital Fountain before joining the faculty of the Massachusetts Institute of Technology. His research interests include computational imaging, sampling, quantization, and source coding theory. Dr. Goyal received the John Briggs Memorial Award at the University of Iowa for the top undergraduate across all colleges. He received M.S. and Ph.D. degrees in electrical engineering from the University of California, Berkeley, where he received the Eliahu Jury Award for outstanding achievement in systems, communications, control, or signal processing. He was awarded the 2002 IEEE Signal Processing Society Magazine Award and an NSF CAREER Award. As a research supervisor, he is co-author of papers that won student best paper awards at the IEEE Data Compression Conference in 2006 and 2011 and IEEE Sensor Array and Multichannel Signal Processing Workshop in 2012. His students have won MIT thesis awards in 2006, 2008, 2010, and 2013. He served on the IEEE Signal Processing Society’s Image and Multiple Dimensional Signal Processing Technical Committee from 2003-09, and he currently serves on the Steering Committee of the IEEE Transactions on Multimedia, on the Editorial Board of Foundations and Trends in Signal Processing, and on the Scientific Advisory Board of the Banff International Research Station for Mathematical Innovation and Discovery. He is Technical Program Committee Chair of IEEE ICIP 2016 and a permanent Conference Co-Chair of the SPIE Wavelets and Sparsity conference series.