Diameter at breast height (DBH) is a fundamental parameter in forest inventory. While traditional manual measurements are labor-intensive, LiDAR-based techniques offer improved accuracy and automation. However, terrestrial and conventional aerial LiDAR systems struggle to efficiently capture DBH due to limited mobility or canopy occlusion. Under-canopy UAVs provide a promising alternative by directly observing tree stems beneath the canopy. This paper proposes a global path planning method tailored for under-canopy UAVs, aiming to improve both safety and DBH estimation accuracy. The approach leverages prior tree distribution to construct distance and visibility fields, generating three-directional observation waypoints around each tree. These points are then globally optimized to ensure safe flight and uniform angular coverage. Simulation and field experiments demonstrate that the planned paths enable complete trunk coverage and effective under-canopy navigation. The extracted DBH values, estimated via least-squares circle fitting, achieve a root mean square error (RMSE) of 6.69 cm. These results confirm the method’s effectiveness in enabling precise and efficient autonomous forest inventory.