Computational photography to describe motion

Bill Freeman

We can use the computer to analyze images, then redisplay them to reveal aspects of the world that we otherwise don't see. I'll present three examples in this theme, all related to motion. (1) Motion without movement can capture, then redisplay, instantaneous visual motion. (2) Motion microscopy analyzes small motions, then re-renders the video with selected motions amplified. (3) Shapetime photography shows the relationship between moving shapes at different times. These projects are joint work with: Ted Adelson, Fredo Durand, David Heeger, Ce Liu, Antonio Torralba, and Hao Zhang.


William T. Freeman is a Professor of Electrical Engineering and Computer Science at MIT. He studied computer vision for his PhD in 1992 from MIT, and has degrees in physics and electrical engineering from Stanford and Cornell. He worked at Mitsubishi Electric Research Labs (MERL) for 9 years before moving to MIT in 2001, and holds 25 patents. His current research focus is machine learning applied to computer vision and computer graphics.