3D Reconstruction Method of Metal Pipe Inner Wall Based on Multi-camera Sequential Image Acquisition
摘要
This paper presents a system and solution for image acquisition and 3D reconstruction of the inner wall of metal pipes. The difficulties of multi-camera illumination and image acquisition of the inner wall of the pipes are solved, the feature point extraction and matching of the inner surface image of the pipe are studied. The sparse point cloud and 3D model reconstruction are realized with high quality. Finally, the accuracy of 3D model is measured and analyzed. In the illumination and image acquisition part, the adapting piece is introduced to install multiple cameras. LED strips are sheltered to avoid the appearance of high light areas, forming an improved dark-field illumination; The pose of cameras is calculated using the rotation matrix to fix the input. SIFT algorithm is used to extract and match feature points of sequential images. After triangulation, meshing and texture mapping, better reconstruction result is obtained by our method compared with other algorithms.