Current UAV path planning methods exhibit efficient performance in navigating environments with small obstacles, such as indoor areas and outdoor forests. However, they often encounter challenges when dealing with environments characterized by large obstacles, such as expansive walls and towering structures scenario (ET scenario). These challenges often cause problems of path redundancy or even complete failure in path planning. In this letter, a hierarchical frontier-based path planning framework is proposed to enhance the adaptability of planners to diverse environments. The collected frontiers, processed by a point cloud voxel down-sampling method, are transformed into a spatially uniform point set. An analysis is conducted to explore the relationship among the UAV's current position, the target, and the potential guide point. Following this, a decision function is designed to identify the optimal local guide point from the point set to guide the UAV in path planning. The comparison of methods in the simulation environment demonstrates the outstanding performance of the proposed method, and its feasibility is further validated through real-world experiments.