Abstract

Computer vision is all about acquiring, interpreting and presenting the rich visual world around us. Today, cameras and displays are ubiquitous and the amount of imagery generated is overwhelming. However, despite significant research strides made over the past decades, scene understanding remains a challenging problem. With the current state-of-the-art, it is hard to reconstruct trees waving in the wind, to capture the motion of gymnasts and animals, to understand and recognize materials such as silk, marble and ivory, to recognize faces and track people under harsh lighting or in the presence of background clutter, to visualize long-range scenes through turbulence, to capture night scenes without blur, and to obtain good contrast images in murky water or fog.

© 2009 Optical Society of America