ROAD : Rapid Optical Asteroid Detection

Abstract

Here we present Tilt and the Three-Dimensional Discrete Radon Transform (3DRT), efficient image processing algorithms capable of identifying streaks at the detection limit in video of the night sky. Tilt and the 3DRT are asymptotically optimal algorithms for the blind search problem, which seeks to identify near-Earth asteroids of arbitrary position and velocity using ground-based optical systems. In the process of establishing the optimality of these algorithms, we formalize the blind search streak detection problem and survey several other state of the art algorithms that solve it: synthetic tracking, Fourier volume rendering, and the Approximate Discrete Radon Transform (ADRT). We also discuss the lessons learned from implementing a near-Earth asteroid detection system which demonstrated the 3DRT's capabilities by identifying five satellite streaks.