Aerial LiDAR Buildings Point Clouds Extraction Combining Region Growing and Principal Component Analysis#br#

 For the airborne LiDAR buildings point clouds extraction process, it is difficult to extract buildings point clouds extraction process, it is difficult to extract buildings points adjacent to the trees, and extraction efficiency of the algorithm after filtering is inefficiency. Aiming at those problems, this paper propose an airborns LiDAR buildings points extraction method based on region growing and principal component analysis. Firstly, the proposed altithom uses airborne LiDAR discrete points which are eliminated gross error to construct the TIN triangulation. After that extracting buildings boundary points according to the feature of the boundary triangle. Secondly, the buildings point clouds would be obtained by region growing with seed points which were from buildings boundary points optimized by neighborhood spatial feature. Finally, the extraction results are checked by principal component analysis algorithm, meanwhile the non-buildings point clouds would be filtered out. Point clouds regions are segmented based on building connectivity and small regions are removed, the final buildings laser foot data is obtained. Three typical LiDAR point clouds data, which are provided by the International Society for Photogrammetry and Remote Sensing, are selected for buildings extraction experiments. By comparing with the traditional morphology and region growing point clouds extraction methods, the test results show that the proposed algorithm can achieve high precision buildings point clouds extraction and have good adaptability to terrain and buildings of different roof types.The reliability of the algorithm is verified.