| Speaker | Pravin Bhat |
| Date | October 6, 2006 |
| Time | 3:00PM to 4:00PM |
| Place | GRAIL (CSE 291) |
Consumer video suffers from low resolution, noise (particularly in low light), and a lack of ability to control the lighting (e.g., no ability to use fash), problems that have been largely overcome in still photography. In this presentation I will discuss our ongoing work on automatically enhancing a video using a few photographs of the scene. For dynamic objects in the video we use an example based method to sample appropriate portions of the photographs in order to enhance the video. For objects in the video that aren't moving we consider the use of multi-view stereo for performing video enhancement.