3D Reconstruction Method Based on Autonomous Attitude Estimation for UAVs

Traditional 3D reconstruction methods usually use the Structure from Motion (SFM) method as the front-end technology, but the SFM usually needs to perform 3D reconstruction after all image data has been collected, which results in a large amount of calculation and a slow speed. The Simultaneous Localization and Mapping (SLAM) method pays more attention to real-time and immediacy, and it can locate and build maps while collecting data in real time. Therefore, this paper proposes a 3D reconstruction method based on visual SLAM for Unmanned Aerial Vehicles (UAVs). In this method, visual SLAM calculates the sparse point cloud map and the attitude information of the UAV. Then using stereo matching, surface reconstruction, and texture reconstruction to perform refined 3D model reconstruction. Based on the above algorithm, this paper designs and implements a system for autonomous positioning and fine mapping of UAVs, and combines it with mixed reality technology to improve the visualization effect of the mapping results. Finally, the effectiveness of the system is verified by experiments.