A method for segmenting the top surface of buildings based on airborne lidar point cloud

Abstract

The present invention proposes a method for segmenting the roof of a building based on an airborne LiDAR point cloud, which achieves high-precision extraction of the roof of a house, and has strong robustness for different data types. The present invention first performs European clustering on the original building point cloud to segment the point cloud of each building; then performs multi-scale segmentation based on L0 gradient minimization for each building respectively to obtain the initial plane segmentation result. In view of the particularity of the house roof extraction work, we have made targeted improvements to the L0 gradient minimization algorithm, including the application of sorting and region expansion strategies and additional geometric constraints, which have improved the L0 gradient minimization algorithm in the plane Accuracy and Robustness in Segmentation. Finally, the graph-cut-based post-processing optimization algorithm is used to refine the initial results, and at the same time solve the problems of over-segmentation and jagged edges that may exist in the initial results to obtain the final roof surface extraction results.

Description

technical field [0001] The invention relates to a method for extracting the roof surface of a house based on an airborne LiDAR point cloud by using L0 gradient minimization and graph cut global optimization, which can be used for LiDAR point cloud classification and segmentation, three-dimensional building model reconstruction, and the like. Background technique [0002] The 3D reconstruction of buildings is an important research topic in the fields of computer vision, photogrammetry and remote sensing. With the rapid development of the laser detection and ranging (Light Detection and Ranging) system, it is possible to quickly and accurately obtain large-scale urban 3D point clouds, which brings great convenience to the reconstruction of 3D buildings. The existing 3D modeling methods of buildings based on airborne LiDAR point clouds can be roughly divided into two categories: model-driven methods and data-driven methods. The model-driven approach is a top-down approach and ...

AI Technical Summary

Problems solved by technology

Aiming at the problem of roof surface extraction of buildings, there are currently many related researches at home and abroad. However, due to the different density of different data point clouds and the different complexity of houses, the existing methods cannot obtain stable and high-precision roof surface extraction results. Therefore, it is necessary to Design a high-accuracy and robust roof extraction algorithm to further advance building model reconstruction

Images