Octree 3D Visualization Mapping based on Camera Information

Abstract

In this paper, an ORB-SLAM algorithm using RGB-D camera collects data for visualization. The collected point cloud data is first processed using a pass-through filter. Statistical filters are used to remove sparse outliers. Thereafter, downsampling is performed using a voxel filter. Increase the sampling amount appropriately according to the actual situation. After mentioned data processing, Octomap is used to visualize the information in Rviz with services provided in Robot Operating System (ROS) platform to build a three-dimensional map. This visualization approach can be applied to preliminary map construction for robot autonomous navigation and 3D modeling of 3D scanner.