Aiming at the difficulty of planning a safe and reliable global path for unmanned ground vehicles in off-road scenarios, a global path planning method based on terrain data is proposed, which integrates the effects of the main driving risks of off-road scenarios on the passability of unmanned ground vehicles. First, the risk characteristics of off-road scenarios are analyzed with the geography remote sensing technology and 3D off-road raster maps that can characterize the main risk characteristics of off-road scenarios are constructed by using the remote sensing image information of off-road areas and Digital Elevation Map. Secondly, based on the main risk features, the slope contraindication table and the surface attribute contraindication table were constructed. In addition, the passable domain of vehicles was determined and the speed of vehicles in the area was specified, and the cost of the passage of unmanned ground vehicles was comprehensively assessed. This method can effectively avoid driving risks while narrowing the search scope of the algorithm and improving the search efficiency. Finally, based on the evaluated access cost, the raster weight matrix is established, the optimized A* algorithm with integrated cost function is proposed, and the planned path meets the vehicle kinematics through Bezier curve interpolation. The simulation results show that the optimized A* path planning algorithm based on terrain data can take into account the influence of risk characteristics of off-road scenarios, and thus plan global paths that can guarantee the passability of unmanned ground vehicles.