In recent years, research on UAV path planning in known environments has matured. However, the demand for planning techniques in unknown environments is increasing in practical engineering applications, where numerous planning needs arise in unknown environments. In such scenarios, the related technologies remain challenging. This paper introduces an efficient 3D UAV Path Planner for Unknown Environments that utilizes a single-layer visibility graph to efficiently handle not only unknown but also partially known environments. It supplements local obstacle handling by using 2D projection maps, aiming to address the high map update storage and computational complexity issues in traditional methods. Extensive experimental results have been tested in a moderately sized virtual campus simulation environment and show that our method is highly efficient with less computational load and map storage compared to state-of-the-art planning methods.