This is an undergraduate graduation design project, the purpose is to get familiar with point cloud segmentation and lay a foundation for further study in this field. Our project is based on this work, which reproduces the basic ideas of the paper very well. Our code has been updated in terms of flatness detection, refinement process, etc., and a 26-neighbor voxel search algorithm has been introduced.
By setting the point cloud file path and parameters(residuals, visualization, evaluation, timing), we can directly run the main.py file to achieve point cloud segmentation.
input_file = '' # 输入点云文件（txt）地址
settings = Settings(residual=0.01,res_th=0.01,dist_th=0.05) # 阈值设定
obrg_calculation(input_file, settings, draw=True, timing=True, evaluation=True) # 决定程序执行哪些操作
