A Dense Map Construction Method Suitable for GIS Metal Cavity Environment
The reflection phenomenon in the metal cavity environment seriously affects the environment perception ability of the inspection robot,resulting in the large area of holes in the construction map.So,this paper presents a dense map construction method suitable for metal cavity environment.This method es-timates the distance of space points by combining depth image and grayscale image,and inhibits invalid pixels and noise in depth image.The optimized depth information and key frame pose were used to calcu-late the point cloud surface,and the sliding window was used to restrict the size of the surface set to reduce the time cost of point cloud search.Triangulation and least square surface fitting were used to repair the holes in the face set to improve the integrity of each frame.Finally,the point cloud map of robot navigation planning is generated by integrating adjacent surfaces with similar normal lines.The proposed method was verified on the open dataset and the actual metal cavity scene.The results show that the proposed method can effectively repair the point cloud holes,significantly improve the model integrity and meet the applica-tion requirements of real-time SLAM.
metal cavityvisual-SLAMdense mappinghole repairtriangulation
万方数据
维普
