基于双目结构光的轨道车辆关键零部件三维重建算法研究

Research on the 3D Reconstruction Algorithm of Key Orbital Vehicle Components Based on Binocular Structural Light

本文通过对双目标定、立体校正与匹配及多频外差求解相位等涉及三维重建的算法进行深入研究,搭建一套基于双目结构光的三维测量系统。该系统不仅仅是增加视角来采集图像,以此获取更完整的点云信息,更重要的是通过投射不同频率且存在相位差的正弦结构光栅,使图像中各像素点都有唯一相位值,经立体校正后,提升左右像面特征点的匹配精度。相较单目结构光将投影仪做逆相机模型来解算深度信息的方式,双目结构光获取的点云信息精度更高,为视觉检测算法更准确的判断螺栓松动及磨耗件到限等故障类型奠定基础。最终实现由基于三维点云数据的故障分析算法来替代人工,完成日检作业。

In this paper,a three-dimensional measurement system is constructed by deeply studying binocular calibration,stereo correction and matching,and other algorithms based on binocular structural light and phase solving.The system is not just adding perspective to collect images to obtain more complete point cloud information.More importantly,by projection different frequency and phase difference sinine structure grating,make the each pixel in the image has the unique phase value,after stereo correction,improve the matching accuracy of left and right features,compared with monocular structure light projector do reverse camera model to calculate depth information,binocular structure light obtain point cloud information precision,more accurate judgment of the bolt loosening and grinding to limited fault type lay the foundation.Finally,the fault analysis algorithm based on 3 D point cloud data is used to replace the daily inspection work manually.