用神经网络进行散乱点的区域分割

点云的区域分割实质上是根据点的局部几何特性的相似性对点进行分类,利用自组织特征映射神经网络（SOFM）方法可以实现无监督的特征聚类。使用SOFM进行反向工程中点云的区域分割,选用数据点的坐标、法向量六维向量作为SOFM的输入,通过改进SOFM的学习算法,加入输入权和距离权,加速了分割的速度和正确性。利用SOFM方法实现点云分割具有以下优点：不必限定面的类型;用户可以控制分区的个数;可以处理噪声数据,实例运行结果验证了此方法的可行性。

Self-Organizing Feature Map Networks for Segmentation of Point-Cloud

Segmentation of point cloud aims at classifying the point cloud into several subspaces and each one can be fitted to a surface. In this paper, a segmentation using self organizing feature map (SOFM) network was presented. Six dimensional feature vector (3 dimensional coordinate and 3 dimensional normal vector) was taken as input for SOFM. Weighted input and weighted Euclidean distance were adopted in the learning process of SOFM, which improves the speed and exactness of the segmentation. The segmentation using SOFM is robust to noise, and has no limitation for surface type. The method is validated by the real scanned point cloud.