Geological record is an important work in the tunnel construction, and it is difficult for traditional artificial geological sketches to quickly and accurately reflect the geological conditions of the tunnel. A methodology for automatically extracting the rock discontinuities of tunnel face from a point cloud is introduced. Firstly, according to the principle of 3D laser scanner, the 3D point cloud is projected onto the plane by quadratic spherical projection and plane projection, and the relative topological relationship between the point clouds is maintained. Then the Delaunay algorithm is used to reconstruct the surface model of the triangle. Finally, the occurrence of each triangle and the distance from the scanning center are calculated. Rock discontinuities are recognized and classified by the clustering algorithm, and which are displayed in different colors. A case study of the Gangbagou tunnel is presented using the proposed method. The proposed method produces results that recognize the most of rock discontinuities with good effect. The research results provide a method for the rapid recording of tunnel geology.