基于图割的多相机管路三维重建测量

Graph Cut Based 3D Reconstruction and Measurement of Tube Using Multi-camera

复杂管路是航空航天和汽车等产品的重要组成部分,近年来视觉测量技术被应用于管路测量中。针对视觉管路测量中管路三维重建效率低的问题,提出了一种基于图割的多相机管路三维重建测量方法。该方法利用图割算法在视差范围内建立管路中心线能量函数获取管路中心线视差,根据视差确定多视图中的匹配点;然后利用多视图最小二乘法获取管路三维中心线点云,对中心线点云进行加权拟合获得特征点并通过NURBS曲线建立管路三维中心线,以管路中心线和管路外径重建管路并测量。实验结果表明,该方法管路视觉测量平均角度误差达到0.12°,满足工装过程需求,具有重建效率高、操作简便的优点。

Complex tube plays an important role in aerospace and automotive.In recent years,vision measurement has been applied to tube measurement.For the problem of low efficiency of 3D reconstruction of tube in vision measurement,a method of graph cut based 3D reconstruction and measurement of tube using multi-camera is proposed.Firstly,the energy function of the tube centerline is established and the disparity is obtained using the graph cut algorithm.According to the disparity,the matching points in the multi-view are determined.Then,the 3D centerline point cloud of the tube is reconstructed based on the least square method,and feature points are obtained by weighted least squares on the centerline point cloud.Finally,the 3D centerline is established through NURBS curve.The tube is reconstructed with the centerline and the outer diameter.It is obtained from the experimental results that the average error of angle is 0.12°.The method meets the needs in the assembly process and has the advantages of high reconstruction efficiency and simple operation.