基于区域生长法的散乱点云分区方法

Segmentation of scattered point data based on region growing method

为高效精确处理散乱点云数据,改进了区域生长算法。首先分析散乱数据点云的高斯曲率和平均曲率,由二次提取法(即先提取平坦点再由高斯曲率和平均曲率的记号提取其他七种曲面类型)形成初始数据分块;再通过区域生长法使粗略数据分块进一步被提取,得到更小的噪声影响及更精确的区域分块。对于各个种子区域,反复控制生长并重建以使最多的点能拟合到单个面片,并由外部噪点来中断循环。实例验证表明该方法具有较强的可操作性和实用性。

To deal with scattered measured points effectively and exactly, an improved algorithm of region growing was presented. The method estimated the Gaussian and mean curvatures of scattered point cloud data. To improve the efficiency of coarse segmentation and provide an initial segmentation, the plane points were extracted, and then seven surface types were extracted from the signs of Gaussian and mean curvatures. Therefore, cloud data was refined by an iterative region growing method. For each seed region, the algorithm iterated between region growing and surface fitting to maximize the number of connected vertices approximated by a single underlying surface, and was terminated by outside noise. Experimental results show the algorithm is maneuverable and practicable.