Design and Verification of a Stereoscopic Imager for Use in Space Close Proximity Operations

Stereoscopic imaging presents an exciting, low-cost alternative to traditional methods of relative navigation for small satellites. This paper discusses the development and testing of a fast, effective computer vision algorithm for such a case. A brief overview of the MR SAT microsatellite and its stereoscopic imaging hard-ware is presented, followed by an in-depth discussion of the software development. The FAST, SURF, and FLANN computer vision algorithms employed in the imaging algorithm are detailed, and methods for reducing navigational errors in the space environment are developed. Results from laboratory testing and Systems Tool Kit (STK) simulations are presented, and imager performance is evaluated.