基于主成分分析的大型管道三维表面重建

Three-dimensional surface reconstruction of large-scale pipe combining principle component analysis

针对大量的点云数据要求快速、方便且精度高的特点,在轮廓线拼接建模方法的基础上提出一种结合主成分分析的三维表面建模方法。由结构光扫描系统获得点云;通过热传导模型将点云进行片层分割并进行主成分分析,得到纬度圈,从而获得纬度圈上的所有对应点对;在相邻点对之间进行柱面插值连接,生成物体的四边形网格模型和三维表面模型。结果显示,被测管道重构表面平均误差为0.19 mm,小于采用基于轮廓线拼接方法得到模型的平均误差。结果表明,采用基于主成分分析的方法进行管道三维表面重构,其精度得到了有效地提高,满足工程测量的精度要求;同时也证明了该方法的正确性。该重构方法适用于表面是一阶连续性被测物体尺寸及变形的分析。

In order to meet the requirements of high accuracy, high processing speed and conveniences for point data, a three-dimensional surface reconstruction is presented combining principal components analysis based on contours tiling. Using the heat conduction model, point cloud from sectional measurement acquired by structured light measuring system is divided into the object layers segmentation, from which the pair-points on adjacent latitude circles are obtained using principal component analysis. Finally, the object model through interpolating cylindrically between adjacent points on the quadrilateral grid connection is reconstructed. Experimental results indicate that the average reconstruction error of the surface model for tube object is 0.19 mm, less than 0.32 mm using the present contour tiling method. The reconstruction accuracy of the 3D surface model is improved and satisfies the requirements of engineering measurement. The method can be applied to estimate the size and deformation of the objects with the surfaces with C1 continuity.