Exploiting Spatial-Angular Temporal Tradeoffs in Photography: Amit Agrawal, Mitsubishi Electric Research Labs (IVC Seminar)

Starts:
2:00 pm on Wednesday, September 25, 2013
Ends:
3:00 pm on Wednesday, September 25, 2013
Location:
MCS 148
Abstract: Conventional cameras have exploited tradeoffs between shutter speed, lens aperture and ISO for photographing a wide range of scenes with low light, fast moving objects and large depth variations. As the sensor resolution keeps on increasing, computational cameras allow trading off pixel resolution for achieving effects that go beyond conventional cameras. In this talk, I will describe several computational cameras that tradeoff pixel resolution for effects such as increasing camera frame rate and capturing light fields. I will show how to capture, model and render wide-angle light fields using low cost mirrors and refractive elements. Finally, I will demonstrate post-capture control of spatial, angular and temporal resolution that allows simultaneously obtaining image, video and light field information for different parts of the scene from a single photo. Bio: Amit Agrawal is a Principal Member of Research Staff at Mitsubishi Electric Research Labs (MERL) at Cambridge, USA. He received his B. Tech in Electrical Engineering from Indian Institute of Technology (IIT), Kanpur in 2000, and MS and PhD from University of Maryland, College Park, in 2003 and 2006 respectively. During 2000-2001, he worked as a DSP engineer at Hughes Software Systems, India. His research is focused in the areas of computer vision and computational photography with emphasis on developing novel cameras and algorithms for scene interpretation, and designing physics based models for vision. He has co-authored more than 30 papers in computer vision and computer graphics conferences including ECCV, CVPR, ICCV, and SIGGRAPH and holds several US patents. He has taught several courses in ICCV and CVPR and has served on the program committee for ICCP, WACV, OMNIVIS and PROCAMS.