Abstract: To improve the efficiency and reliability of plane segmentation for point cloud simultaneously,a novel segmentation approach with the combination of region growing and RANSAC is presented.For the automatic selection of the seed plane,the octree nodes are verified for their flatness,the parameters of which are using as the region growing constraints to get the initial segmentation results. Then the nodes are checked for the need of further division and segmentation according to the initial results.The final results are gained after the post processing of merging and refining parameters using RANSAC.During the segmentation process,the computation of the points’ properties such as normal vector is not required. The implemented experiment shows that our method is efficient and robust.