Sampling for Large-scale Image Restoration

Event Date: April 24, 2014
Speaker: Dr. Ho (Stanley) Chan
Speaker Affiliation: Post-Doctoral Research Fellow, Harvard University
Sponsor: Potential ECE faculty candidate
Time: 10:30am
Location: EE 317
Contact Name: Professor Charles Bouman
Contact Phone: 765-494-0340
Contact Email: bouman@purdue.edu

Abstract:

Over the past decade, the astonishing evolution of devices and materials has altered the landscape of imaging technologies --- both in terms of quantity (e.g., the ubiquitous hand-held cameras) and in terms of quality (e.g., the ever growing resolution). To further reduce pixel pitch and miniaturize imaging systems, boost frame-rates, and capture videos in harsh, low-light scenarios, our ability to restore images at large scales becomes a fundamental factor. By advancing large-scale image restoration, we can open the door to a new generation of computational imaging, for applications to physical science, robotics, health care, and beyond.

In this talk, I will present a sampling framework for restoring large-scale images. The new framework allows us to effectively restore images using a small subset of samples, hence significantly reduces the computational complexity. Three analytical questions will be discussed: (1) Is there any statistical guarantee of the sampling framework? (2) What is the optimal sampling strategy? (3) How do we achieve optimal sampling in practice? I will demonstrate applications to single-image denoising and denoising using external databases.

Short Bio:

Ho (Stanley) Chan is a post-doctoral research fellow in the School of Engineering and Applied Sciences at Harvard University. He received the B.Eng. degree in Electrical Engineering (with first class honor) from the University of Hong Kong in 2007, the M.A. degree in Mathematics from University of California, San Diego in 2009, and the Ph.D. degree in Electrical Engineering from University of California, San Diego in 2011. His current research interests include statistical signal processing and graph theory, with applications to imaging and network analysis. Dr. Chan is a recipient of the Croucher Foundation Fellowship for Post-doctoral Research 2012-2013 and the Croucher Foundation Scholarship for PhD Studies 2008-2010.