New York University
Department of Physics
Back to Previous Page
spacer
Weekly Bulletin 12-day Event Forecast All Scheduled Events
Astrophysics and Relativity Seminars Atomic Molecular and Optical Seminars High Energy Physics Seminars
Condensed Matter Physics Seminars Experimental Particle Physics Seminars Other CCPP
Physics Colloquia Other Events Events Search

Weekly Bulletin

May, 05/24/2013
Events and times subject to change

MON
05/20
TUE
05/21
WED
05/22
THU
05/23
FRI
05/24
MON, 05/20 (Back to top)
May 20, 2013 Monday 12:30 PM 
Meyer 5th Fl. CCPP Lounge
Other CCPP (ccpp)

CCPP Brown Bag
Dan Zwanziger
NYU

Confinement: scenario or theory?



TUE, 05/21 (Back to top)
May 21, 2013 Tuesday 11:00 AM 
Meyer 5th Fl. CCPP Lounge
Other CCPP (ccpp)

AstroCoffee

informal discussion of recent astro papers


May 21, 2013 Tuesday 2:00 PM 
Meyer 6th Floor Conference Room
Soft Condensed Matter Seminars (csmr)


Jerome Fung
Harvard University

TBA



WED, 05/22 (Back to top)
May 22, 2013 Wednesday 3:00 PM 
Meyer 6th Floor Conference Room
Other Physics Department Events (other)

Physics Department Graduation Party

Light refreshments will be served.
RSVP to Lorelei DeMesa at 212-998-7711 or via e-mail ljd5@nyu.edu.


FRI, 05/24 (Back to top)
May 24, 2013 Friday 11:00 AM 
Meyer 5th Fl. CCPP Lounge
Other CCPP (ccpp)

AstroCoffee

informal discussion of recent astro papers


May 24, 2013 Friday 2:30 PM 
719 Broadway, Room 1221
Other CCPP (ccpp)

VLG Seminar
Bill Freeman
MIT

Re-rendering motions: motion denoising and motion magnification

I'll describe two projects at the intersection of vision and graphics. (1) Motion denoising processes a video to remove the flickering motions common in time-lapse sequences, while revealing the long term changes. We define a cost function for rearrangments of the pixels over time and space to favor the desired processing, and use loopy belief propagation to find an approximate solution. (2) Motion magnification amplifies small motions to make them more visible. For this we use a multi-scale signal processing approach, which can be applied in real time. The magnified motions can reveal "a big world of tiny motions", showing properties of the world not otherwise visible. Both two techniques can be applied to videos of any temporal sampling rate, but motion denoising is best suited to timelapse sequences, while motion magnification matches works best for higher sampling rates (ordinary or high-speed videos).