Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision
(CS 528)

High-performance imaging using dense camera arrays

Marc Levoy
Computer Science Department
Stanford University Monday, Sept. 29, 2003, 4:15PM
TCSeq 200


The advent of inexpensive digital image sensors, combined with the decreasing costs of computing and memory, have led to widespread interest in building sensing systems that incorporate large numbers of cameras. To explore this area, the Stanford Computer Graphics Laboratory has designed a scalable architecture for arrays of miniature video cameras. Our current implementation contains 128 custom, CMOS-based cameras, each of which measures 30mm on a side. The system is designed to record 128 synchronized, slightly compressed video streams through four PCs to a striped disk array.

Multi-camera systems can function in many ways, depending on the spacing and arrangement of the cameras. If the cameras are packed closely together, then the system functions as a single-center-of-projection synthetic camera, which we can configure to provide high performance along one or more imaging dimensions, such as spatial resolution, signal-to-noise ratio, dynamic range, depth of field, frame rate, or spectral sensitivity. If the cameras are placed farther apart, then the system functions as a multiple-center-of-projection camera, and the data it captures is called a (video) light field. If the cameras are placed at an intermediate spacing, then the system functions as a single camera with a large synthetic aperture. Such a system has the ability to focus narrowly on one depth plane, making objects in front or behind that plane so blurry that they almost disappear. This allows us to see through partially occluding environments, like foliage or crowds. In this talk, I will describe the architecture of our system, the hardware of our current 128-camera implementation, and some of the applications we are developing for it.

This is joint work with Mark Horowitz and several of our PhD students. The architect of the camera array is Bennett Wilburn.

About the Speaker

Marc Levoy is an Associate Professor of Computer Science and (jointly) Electrical Engineering at Stanford University. He received a Bachelor's and Master's in Architecture from Cornell University in 1976 and 1978, and a PhD in Computer Science from the University of North Carolina at Chapel Hill in 1989. In the 1970's Levoy worked on computer animation, developing an early computer-assisted cartoon animation system. In the 1980's Levoy worked on volume rendering, a family of techniques for displaying sampled three-dimensional functions such as medical scanner data. In the 1990's he worked on technology and algorithms for digitizing three-dimensional objects. His current interests include sensing and display technologies, image-based modeling and rendering, and applications of computer graphics in art history, preservation, restoration, and archaeology. Levoy received the NSF Presidential Young Investigator Award in 1991 and the SIGGRAPH Computer Graphics Achievement Award in 1996 for his work in volume rendering.


Back to the Colloquium Page