3D Reconstruction of Flexible Cables Based on Binocular Vision

In the process of robot-assisted automatic assembly of flexible cables, to better complete the assembly task, visual assistance through cameras is required. This project proposes a method using the watershed threshold segmentation algorithm to extract flexible cables for stereo matching, addressing the issue of high mismatch rates in stereo matching caused by weak textures and reflective surfaces of flexible cables. This lays a solid foundation for the automatic robotic assembly of flexible cables. Firstly, the Zhang's calibration method is employed to conduct calibration experiments for the binocular camera. Subsequently, the calibration parameters are utilized to perform stereo rectification on the captured cable images. After obtaining the rectified images, the watershed segmentation algorithm is applied to extract the cables. Then, stereo matching is conducted to obtain the three-dimensional coordinate table of the cables. Finally, the PCL (Point Cloud Library) is used to reconstruct the three-dimensional model of the cable. Experimental results demonstrate that this cable extraction method can rapidly establish a three-dimensional model of flexible cables and provide users with the pose information of flexible cables through a point cloud visualization window.