Srinivasa Narasimhan (Robotics Institute Carnegie Mellon University)
Tuesday, 12.5.2009, 11:30
Light sources and cameras are optical duals: sources emit light rays while the cameras capture them. This talk will argue that light sources can serve as better cameras in many applications, advancing the state of the art in computer vision. (a) By moving a light source instead of a camera, we show how to reconstruct highly intricate shapes like wreaths, corals and tree branches from hundreds of 'views'. (b) We leverage the 'illumination dithering' in the micromirror array of DLP projectors to speedup virtually any active vision algorithm, resulting in high-speed 3D reconstruction, photometric stereo, appearance capture and high frequency preserving motion blur. (c) We exploit high spatial frequency and shallow depth-of-field of projectors to recover scene properties despite complex light propagation through materials. (d) Finally, we investigate what the main outdoor illuminants (sky and sun) can tell us about the camera and how to recover outdoor illumination from webcam sequences and consumer-grade photographs.