There is an enormous amount of digital imagery available today, but at present
there is no good way to search it. I would like to build a search engine for
imagery (like AltaVista, but for pictures and videos). My research focuses on
algorithms for extracting content from images, which is the most challenging
part of this project.
I will describe new methods for three classical problems of
content-extraction. The first problem involves searching a database of static
images for the most similar image to a given query. The second problem is to
segment a video sequence into scenes by detecting scene breaks (such as cuts
and dissolves). The third problem is to detect moving objects in a video
sequence. I will also provide a few underlying themes which appear in all
these algorithms.
This is joint work with Kevin Mai, Justin Miller and Greg Pass.
BIOGRAPHY:
Ramin Zabih is an assistant professor in the Computer Science Department at
Cornell University. He received S.B. degrees in Computer Science and in Math,
and an M.Sc. degree in Computer Science, from the Massachusetts Institute of
Technology. His Ph.D. is in Computer Science from Stanford University. His
research interests are in computer vision and multimedia. He has served as a
consultant for several companies, including Xerox PARC and Interval Research.