Terrain Relative Navigation (TRN) has enabled two successful rover touchdowns on Mars and a sample recovery mission on the asteroid Bennu. Many future missions plan to use TRN including vehicles for NASAs Commercial Lunar Payload Services (CLPS) and the Human Lander System (HLS). A TRN algorithm identifies surface features in images from spacecraft cameras and matches them to features in other images or reference maps. Those matches are processed to determine the spacecraft’s motion. Consistently matching features in a scene is difficult under different lighting conditions and from different camera positions. Past missions required significant human supervision to prepare surface maps or ensure trajectories were illuminated well. This research proposes an approach to identify feature properties which don’t change (are invariant) to different illumination and viewing conditions. Reflectance models for celestial bodies have been primarily developed for planetary science applications. This work will quantify the variance in feature matching performance due to perturbations in illumination model parameters. From this a reflectance model containing the parameters contribute the most to TRN. The existence of illumination invariants will be evaluated using the proposed model. An illumination invariant descriptor (a fingerprint) will be constructed from the invariants identified. The descriptor will be tested to ensure it can be identified and located accurately when model parameters are varied. Finally, the feature descriptor will be tested under two scenarios. The first scenario will quantify the descriptor’s feature matching performance under ideal conditions using synthetic imagery generated using the TRN reflectance model. The second scenario feature matching performance on real imagery from the OSIRIS-REx mission to Bennu. The resulting technology from this research will include a TRN reflectance model and an illumination invariant feature descriptor. This research will also demonstrate the ability to increase spacecraft autonomy using an illumination invariant feature descriptor for TRN.