由于3维扫描点云通常存在噪音和缺失数据，提出了一种鲁棒的点云网格重建算法。对张量矩阵方法估计的点云法向进行增强特征处理，在频域中进行3维快速傅里叶变换，提取粗糙离散等值面。原始点云经梯度方向迭代移动后，过滤噪音和剔除离群点，并修补点云缺失数据。点云被自适应筛选后，利用圆球相交的方法生成新的三角形。实验表明，该算法具有快速、稳定可靠和内存消耗小的优点。

There is noise and defective data on the 3D scanning point cloud. A robust mesh reconstruction algorithm is proposed. Surface normals are estimated by tensor matrix with enhanced features. By computing 3D fast Fourier transform (FFT), discrete iso-surface is extracted. Points are moved onto the iso-surface by an iterative clustering along gradient field, where the noise and outliers are removed and defective data are repaired. Point cloud is decimated adaptively, and then a new triangle is generated using sphere-intersected method. The experimental results have shown that the algorithm is fast, robust and use low memory.