基于实际双目相机的点云三维重建算法研究

Stereo vision is one of the difficulties in the field of computer vision. Two images of the target object are obtained by using cameras at different positions, and the parallax between corresponding feature points in the images is calculated to obtain three-dimensional information. A series of processes of stereo vision are required to use the actual stereo camera for the 3D reconstruction of visual simultaneous localization and mapping (SLAM). Therefore, a dense point cloud 3D reconstruction algorithm based on the actual binocular camera is designed in this paper. It mainly includes the calibration of stereo camera, stereo image correction and stereo matching algorithm, dense point cloud 3D reconstruction and point cloud optimization and management. The experimental results show that the proposed algorithm can successfully apply the large baseline stereo camera to the actual indoor and outdoor scenes to conduct dense 3D reconstruction of the environment, so as to improve the real environment information while positioning, and verify the practicability and effectiveness of the proposed algorithm, which can be provided to the robot or automatic driving to facilitate automatic navigation.