Abstract:Aiming at the problems of low accuracy and poor adaptability in extracting building roof using LiDAR point cloud data, we propose a stepwise method for high-precision extraction of building roof point clouds. We calculate the reliability index of the point cloud through principal component analysis, select the reliable plane points, then use the K-means algorithm to realize the clustering of the reliable points in the normal vector space and extract the initial roof patch through stepwise plane estimation. Finally, we process the combination of building roof patches and the attribution judgment of unmarked points. The test results show that the proposed method has excellent extraction results, high extraction efficiency, and can obtain better extraction results for building roofs of different complexity levels.