Current generation of panoramic imaging systems are based on three types of technologies: 1) Catadioptric sensors, as in omni-directional cameras, using combination of lenses and mirrors in a carefully arranged configuration relative to a standard camera; 2) Alignment of single-line scans or strips as a single camera rotates; 3) Alignment of images from multiple cameras with negligible baselines relative to scene distance, each camera covering a small section of the entire view. This project explores the multi-camera design to achieve super-resolution panoramic motion imagery for UAV deployment. Among several attractive advantages, including simplicity, flexibility, reconfigurability, and robustness of the design, this is a solution based on a low-cost technology. A detailed investigation of various design parameters and how they impact system performance (accuracy, resolution, and achievable frame rate) for target detection and tracking, visually-guided positioning, and terrain mapping from UAVs are investigated. The deliverables comprise performance graphs and charts where critical system parameters can be identified, configurations most suitable for targeted applications at multitude of operational altitudes can be chosen based on quantitative measures, and guidelines are established for system design and construction in the 2nd phase, which also involves the deployment for collecting real data (onboard suitable UAVs).
