Accurate Planetary Limb Localization for Image-Based Spacecraft Navigation

The use of images for spacecraft navigation is well established. Although these images have traditionally been processed by a human analyst on Earth, a variety of recent advancements have led to an increased interest in autonomous imaged-based spacecraft navigation. This work presents a comprehensive treatment of the techniques required to navigate using the lit limb of an ellipsoidal body (usually a planet or moon) in an image. New observations are made regarding the effect of surface albedo and terrain on navigation performance. Furthermore, study of this problem led to a new subpixel edge localization algorithm using Zernike moments, which is found to outperform existing methods for accurately finding the horizon's location in an image. The new limb localization technique is discussed in detail, along with extensive comparisons with alternative approaches. Theoretical results are validated through a variety of numerical examples.