复杂工况下的多传感融合点云空洞互补方法

Key Points

• This research aims to address point cloud gaps in 3D underground reconstruction caused by complex environmental factors.
• Combined calibration of laser radar and depth camera for consistency in time-space relationships.
• Used residual regression for noise reduction in depth camera point clouds.
• Evaluated slice continuity with weighted spectral entropy (WSE).
• Dynamically selected optimal slices for adaptive fusion to fill gaps.
• Outperformed comparison methods in Chamfer Distance (CD) and Earth Mover's Distance (EMD) under various lighting conditions.
• Reduced CD by up to 85.82% and EMD by up to 90.68%.
• Achieved size errors of less than 20 mm.