TAMU Computer Science: Panoramic vision

The objective of this team was to develop a vision system that could capture 360-degree video at (close to) frame rates. Rather than using a catadioptric system (i.e., camera and mirrors), the team used an array of webcams arranged in a ring configuration. Cameras were calibrated off-line, and their streaming video was "stitched" together in real-time. The project involved a significant reverse-engineering effort for the webcams, as well as computer vision algorithms for image calibration and stitching.