Computational Imaging with Quanta Image Sensors

Speaker: Stanley H. Chan
Affiliation: Purdue University

Abstract: Since the birth CMOS image sensors in the early 90s, pixel pitch has been continuously shrinking in order to increase spatial resolution, lower power consumption, and reduce size. However, as pixel pitch shrinks, the signal-to-noise and dynamic range of the sensor also drops, thus causing fundamental limits. Quanta Image Sensor (QIS) is a possible candidate for the next generation image sensor with primary applications in science and defense, but also potential applications in consumer electronics. Unlike conventional CMOS sensors which accumulate photons, QIS partitions a unit pixel into many tiny cells that only detect the presence or absence of a single photon.  As a result, data acquired by the QIS is a massive random binary bit stream and it requires computational methods to decode the image.  In this presentation, we will discuss two computational imaging problems associated with QIS.  First, we will discuss fast and robust algorithms to reconstruct the image from the binary bit streams. These methods span from convex optimization algorithms to the most recent non-iterative Transform-Denoise algorithm developed at Purdue. Second, we will discuss optimal threshold control schemes to maximize the dynamic range of the sensor. We will prove the existence of a phase transition that will guide the design of a practical algorithm. Towards the end of the presentation we will outline a few open problems to be addressed in the future.

Biography:  Stanley H. Chan is currently an Assistant Professor in the School of Electrical and Computer Engineering and the Department of Statistics at Purdue University, West Lafayette, IN. He received the Ph.D. degree in EE and the M.A. degree in Mathematics from UC San Diego, in 2011 and 2009, respectively, and the B.Eng. degree in EE from the University of Hong Kong in 2007. Prior to joining Purdue, he was a postdoctoral research fellow at Harvard from 2012 to 2014.  He is a recipient of the Best Paper Award of IEEE International Conference on Image Processing 2016, IEEE Signal Processing Cup Second Prize, Purdue College of Engineering Outstanding Graduate Mentor Award, and Eta Kappa Nu Outstanding Teaching Award. He is also a recipient of the Croucher Foundation Fellowship for Postdoctoral Research and the Croucher Foundation Scholarship for PhD Studies, two of the most prestigious scholarships in Hong Kong.  He has been an Associate Editor of the OSA Optics Express since 2016, an Elected Member and the subcommittee Chair of the IEEE Signal Processing Society Special Interest Group in Computational Imaging since 2015. He was the co-chair and co-organizer of the computational imaging special session in ICIP 2016, and a technical committee member of ICIP, ICASSP, OSA Imaging and Applied Optics Congress, and Midwest Machine Learning Symposium.

For more information, contact Prof. Christina Fragouli (christina.fragouli@ucla.edu)

Date/Time:
Date(s) - Nov 20, 2017
2:00 pm - 3:30 pm

Location:
E-IV Faraday Room #67-124
420 Westwood Plaza - 6th Flr., Los Angeles CA 90095