For the future exploration of our solar system, missions to Small Solar System Bodies (SSSBs), such as asteroids or comets, are a promising and scientifically important area. For navigation and guidance tasks, high-resolution and high-accuracy 3D distance maps of the surface are required. Flash-LiDAR sensors are often used to capture the surface of a SSSB, but they are limited in resolution. This paper proposes an artificial intelligence (AI) based approach to simultaneously improve the resolution and accuracy of 3D distance maps generated by flash-LiDAR through AI-based data fusion with 2D camera images. Our method leverages the smaller ground sampling distance (GSD) of the 2D camera images and the distinct error behavior of 3D surface reconstruction from LiDAR and camera data. A generative adversarial network (GAN) architecture was designed for that purpose. We evaluated our approach extensively with a custom generated synthetically dataset of an asteroid surface. As a result of the tests, a simultaneous improvement of the ground resolution by a factor of 4 × 4 and suppression of the RMSE distance error by a factor of 1.31 with respect to the simulated flash-LiDAR data was successfully demonstrated.