Interval-Based Celestial Geolocation Using a Camera Array

To replicate classical celestial navigation algorithms with modern technologies, a hemispherical camera array is used to capture panoramic images of the sun for the purpose of obtaining azimuth and zenith angles. A novel geolocation algorithm is presented to use these angles to determine the absolute location without aid of satellites. It derives both the geolocation estimates and the error intervals based on measurement noise levels. Hough Transform, charge-coupled device blooming, and interval analysis are employed to improve accuracy even with lower quality sensors. The proofs of our theoretical results based on an interval analysis handle arbitrary celestial references. Calibration of and geolocation from our experimental prototype demonstrates the feasibility and potential of our system. We obtain geolocation to within a rectangular area of 0.533 km × 1.6 km using an array of inexpensive cameras with 640 × 480 pixels resolution. Our simulation result shows that the improvement of sensor precision with technology already available can reduce the area of uncertainty to less than 200 m².