Bullet Detection and Trajectory Estimation Using a Single Camera

Abstract

Closed-loop fire control systems greatly increase a weapon system's ability to success- fully engage a target by measuring the trajectories of its outgoing projectiles. This has been demonstrated with both line-of-sight and non-line-of-sight systems, but research in autonomous Remote Weapon Stations (RWS's) currently lacks this capability. One of the major challenges in bringing this capability to RWS's is that radar is not seen as an appro- priate sensor due to its size, cost, and active nature. The goal of this research was to determine if projectile trajectories could be accurately measured with a camera. Algorithms were developed to detect small-caliber tracer rounds in images and then combine those detections with a ballistic model to estimate trajectories. These algorithms were successfully tested with live gunfire data, which showed that tracer rounds were readily detectable and that estimated trajectories accurately predicted impact points. Simulations also show that the algorithms can distinguish individual tracer trajec- tories from machine gunfire and eliminate several types of outliers. These results open the possibility of providing closed-loop fire-control to RWS's by using a camera as the main sensor.