Onboard Image-Processing Algorithm for a Spacecraft Optical Navigation Sensor System

A new image-processing algorithm is presented that is capable of autonomously extracting optical navigation measurements front raw images taken by a spacecraft in the vicinity of a planet or moon. The algorithm is designed to support autonomous navigation onboard a spacecraft without Earth communication. The planet or moon is modeled as a triaxial ellipsoid and the two-dimensional image is fitted to the three-dimensional model using conic section curve fitting methods. The performance of this algorithm is demonstrated with real images of Earth, the moon, Venus, Mercury, and Phobos taken by spacecraft under different operating conditions. Analytic methods are also presented for computing the corresponding measurement covariance matrices and measurement sensitivity matrices.