Video Magnification

We are developing algorithms to manipulate temporal variations in videos, to reveal small imperceptible changes (check out this video, and read more about it here) as well as automatically remove distracting changes (see here).

We are looking for a strong and motivated student to work closely with us on exploring potential applications. This includes trying out our methods with different kinds of data, such as medical images (fMRI), satellite imagery, seismic data and time-lapse videos, developing tools to facilitate the experiments, and potentially helping us tune and improve the algorithms.

Prerequisites: good MATLAB/C++ programming skills, experience in computer vision and/or image processing is a plus.

Contact: Michael Rubinstein (mrub@mit.edu), Fredo Durand (fredo@mit.edu), Bill Freeman (billf@mit.edu)