Abstract: For the detection of lava tubes on extraterrestrial bodies, this paper utilizes LiDAR technology to conduct field surveys on lava tubes in the Shishan Volcanic Cluster area of Haikou City, which has a high degree of similarity with the Moon and Mars. It also proposes a skylight detection method based on grid division and applies a crack detection method based on data enhancement and multi-scale feature learning network to detect the internal structure of the lava tube and study its morphology. The interior of the lave tubes is scanned in all directions using a portable handheld laser scanner in the first. Based on the high-precision point cloud data obtained by measurement, this paper completes a three-dimensional model using the ContextCapture software, and some special miniature geomorphic like skylight in the tube and the tubes’ inner fissure is detected and extracted automatically. The experimental results show that the laser scanner combined with point cloud processing software and program can efficiently, intuitively and truly restore the morphology and structure of the lava tube, conduct quantitative analysis, and detect the special miniature geomorphology in the tube.