基于单边雅可比方法的点云平面拟合

Plane fitting of point clouds based on unilateral Jacobi method

针对三维激光雷达点云数据中的异常值影响平面拟合精度的问题,提出了一种基于单边雅可比变换的点云平面拟合方法。该方法首先利用半径滤波法对点云数据进行预处理以剔除大部分噪声点,然后以单边雅可比旋转法对点云坐标数据构成的数值矩阵进行正交变换与归一化,得出奇异值分解,进而得出拟合平面的参数。通过仿真,将该方法与随机采样一致性方法、特征值法进行对比实验,结果表明该方法在抗噪声点干扰方面具有明显优势。

To overcome the problem that the abnormal points in 3D LIDAR point clouds will affect the plane fitting accuracy,a method for plane fitting of point clouds based on unilateral Jacobi method was proposed.In this method,radius filtering was firstly used to pre-process the point clouds,and then the unilateral Jacobi method was used to conduct orthogonal transformation and normalization on the numerical matrix composed by the point cloud coordinates,so as to obtain the result of singular value decomposition,and the feature vector corresponding to the smallest singular value is the coefficient vector of the fitting plane.Compared with the random sample sonsensus(RANSAC)algorithm and eigenvalue method,the results show that the proposed method has obvious advantages in anti-noise point interference.