Wednesday, January 31, Volen 101, 2:00-3.00 pm
In this talk, I will present current research in the rapidly developing field of image-based rendering. Image-based rendering describes a new class of computer graphics techniques where the underlying scene representations are described by a collection of images rather than models of surface geometry and material properties. Image-based rendering fundamentally recast the computational approach to image synthesis from a problem of simulation to a problem of signal reconstruction.
I will discuss two methods for synthesizing novel views of a scene given some set of images. The first method does not rely on any underlying structural model or image correspondences to synthesize new views. Instead, it treats a collection of cameras as a synthetic-aperture camera array with a natural and intuitive user interface. The second method that I will describe uses an approximate geometric representation that can be rapidly acquired and computed. When this representation is combined with images, it achieves results that are comparable to an exact geometric model. This method employs, what is among the simplest and most robust of computer vision techniques. Finally, I will discuss hybrid systems in which approximate structural models describe parts of an environment while the remaining scene elements are models using no structural description.
I will conclude with a discussion of the many opportunities presented by image-based rendering for new synergies between the fields of computer graphics and computer vision. I will suggest how many of the intermediate results from computer vision (intermediate in the sense that they do not directly result in a description of three-dimensional structural) can be directly applied to the problems of rendering novel viewpoints. Furthermore, I will discuss how new image-based rendering techniques provide a new set of tools for the efficient visualization and exploration of computer vision results.
Biography:
Leonard McMillan is an Assistant Professor in the Electrical Engineering and
Computer Science department at the Massachusetts Institute of Technology.
He is a member of the MIT Laboratory for Computer Science and a co-leader of
the MIT Computer Graphics Group. Leonard received his Bachelors (’83) and
Masters (’84) degrees in Electrical Engineering from Georgia Institute of
Technology. Leonard received his Ph.D. in Computer Science from the
University of North Carolina at Chapel Hill (’97). Leonard has been a Member
of Technical Staff at AT&T Bell Laboratories where he worked in the Digital
Signal Processing Architecture Group and was a co-architect of the AT&T
Pixel Machine. Leonard has also worked as a Senior Staff Engineer at Sun
Microsystems where he helped develop several visualization and multimedia
products.
Leonard is a pioneer in the field of image-based rendering. Image-based
rendering is a new approach to computer graphics where scenes are rendered
directly from a collection of reference images rather than a geometric
model. Leonard is also interested in a wide range of related topics
including computer graphics rendering, imaging methods and technologies,
three-dimension display technologies, computer graphics hardware, and the
fusion of image processing, multimedia, and computer graphics.
Host: Liuba Shrira