Lunar Image Matching Based on Impact Crater Morphology and Distribution for Landing Localization

Lunar image matching is often employed to locate a lander by aligning the lander’s descent images with orbiter images that possess geographic coordinates. However, repetitive lunar topographical features and variations in illumination introduce challenges to this process. This study proposes an innovative method leveraging the distribution characteristics of impact craters. Our method encompasses three steps: (1) detecting impact craters from images to obtain their centers and radiuses; (2) constructing rotation and scale invariant crater feature histogram descriptors based on the impact craters; (3) applying RANSAC to eliminate outliers and achieve precise matching. The research focuses on the landing area of Chang'e-5, utilizing data from its descent camera and the Lunar Reconnaissance Orbiter Camera Narrow Angle Camera (LROC NAC) for empirical analysis. A comparative evaluation with the traditional SIFT algorithm demonstrates the superior effectiveness of the proposed method, yielding a registration error of approximately 8 pixels in the reference image.