Computer Science Thesis Proposal

— 1:30pm

In Person - Reddy Conference Room, Gates Hillman 4405

DORIAN CHAN, Ph.D. Student, Computer Science Department, Carnegie Mellon University

Diffraction and Computer Vision

Historically, computer vision systems struggle with sensor limits, bandwidth usage and power draw when applied in the wild, limiting their practicality. In this thesis, we explore how light diffraction can be used to mitigate these challenges in real applications. Thanks to the unique coding of light that diffractive optics provide, we demonstrate compressive high-speed imaging that better encodes scene content into available camera bandwidth compared to past approaches. Leveraging the unique image formation model of diffracting light, we implement high-dynamic range illumination systems that enable faster, more programmable structured light systems and longer range time-of-flight cameras without increasing power usage. With these ideas in mind, we propose a technique for sensing the fine vibrations of objects over a large field-of-view via diffraction, an approach for probing light transport based on programming the coherence of a projector system, and a methodology for spatially controlling camera resolution by using a programmable diffraction grating. 

Thesis Committee: 

Matthew O'Toole (Chair)
Ioannis Gkioulekas
Aswin Sankaranarayanan
Mohit Gupta (University of Wisconsin-Madison)

Additional Information

Add event to Google
Add event to iCal