In order to solve the problem that point clouds of intersecting and crossing lines may be difficult to deal with, insulator point clouds are difficult to be segmented, and the existing algorithm models are incomplete, fast extraction and reconstruction method of power line based on point cloud data features was adopted. Firstly, fast rough classification of point clouds was carried out based on voxel grid elevation characteristics of transmission lines. Then point cloud was refined rapidly by using random consistency algorithm. Point cloud of power line was extracted efficiently and accurately by filtering algorithm taking into account cross-line. In view of local linear model of power line, insulators were segmented. Finally, the local weighted centroid method was used to extract the key points of power line. Fast reconstruction of power line was realized. Theoretical analysis and experimental verification were carried out. Good experimental results were obtained. The results show that, overall accuracy of power line extraction and reconstruction is 95.3%. Total time consumed is less than 2.5s. The speed, accuracy and robustness of the algorithm are verified. This result is helpful to extract and reconstruct point cloud of power line.