改进的微分流形曲面重建算法  

Improved differential manifold surface reconstruction algorithm

作　　者：李瑞[1] 韩慧妍[1] 韩燮[1] 

机构地区：[1]中北大学计算机与控制工程学院,山西太原030051

出　　处：《计算机工程与设计》2016年第5期1325-1330,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61379080);国家科技支撑计划基金项目(2013BAH45F02);山西省基金项目(2015021093)

摘　　要：为提高大规模散乱点云的重建精度和效率,在RBF隐式曲面重建算法的基础上提出一种改进算法。减少平凡RBF中心点数,降低方程组求解规模,采用八叉树对待重建的点云进行分割,根据各点主曲率的Hausdorff距离提取模型中的特征点。为从整体上构建模型,避免繁杂的拼接操作,将RBF曲面的规则定义域扩展到微分流形上,为每一个分块点云建立坐标卡,通过求解线性方程组得到模型的权值系数,利用最短距离为控制顶点建立基函数,通过复合流形曲面上的单位分解和控制顶点得到最终模型。实验结果表明,该算法适用于任意拓扑的曲面重建,具有较高的重建效率和精度。To improve the reconstruction efficiency and precision of large scattered point cloud,an improved method based on RBF implicit surface reconstruction was put forward.To reduce the number of center points of ordinary RBF as well as the scale of equation set,the point clouds were segmented using octree,feature points of the model were extracted based on Hausdorff distance of each point.The model was constructed as a whole and the complicated mosaic was avoided,the regular domain of RBF surface was expanded to differential manifolds.Atlas for each point cloud block was established,and the weight coefficients of model were obtained after solving a linear equation set,and base function for control vertices was built using shortest distance.Unit decomposition and control vertices were composited to obtain the final surface model.Experimental results show that the algorithm is applicable to any topology surface reconstruction,and has high precision and efficiency.

关 键 词：点云重建 微分流形 RBF函数 HAUSDORFF距离 隐式曲面 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]