To address the low efficiency and poor accuracy of hidden danger detection in UAV inspections of transmission channels using traditional methods, this paper proposes a 3D point cloud-based method for detecting hidden dangers in transmission channel space, which effectively enhances detection accuracy and efficiency. Firstly, an adaptive density down-sampling algorithm is designed to achieve sparse and density homogenization of point cloud data, and the aerial point cloud and ground object point cloud are separated by an elevation grid algorithm. Secondly, the traditional RANSAC algorithm is improved, and the seed point selection method and inner point determination function are optimized by introducing curvature factor to achieve power line point cloud extraction. Finally, the KD-Tree structure is constructed to determine the hidden danger region and the hidden danger is classified by the method of Angle variance. Through field data collection experiments, the results show that the proposed method can effectively detect the spatial hidden dangers of transmission channels, the accuracy and recall rate of power line point cloud extraction can reach 96.8% and 97.1%, and the accuracy rate of spatial hidden dangers detection can reach more than 96%. Compared with the traditional method, the proposed method has obvious advantages in the accuracy and efficiency of the hidden trouble detection in the transmission channel space, and has good practical value in the intelligent inspection of the transmission channel.