基于点云数据的立木树干局部曲面拟合及拓扑结构

Local surface fitting and topology structure based on timber trunk of point cloud

基于海量测量点云数据加工处理的关键是通过获得点云的局部特征拓扑结构来精简数据,而其算法的效率尤为重要。本研究首先对缺乏足够几何拓扑信息的点云,建立每个数据点邻近点的几何拓扑信息,同时综合运用重构管道曲面和随机霍夫变换算法,对立木树干进行拟合。实验结果表明,其效果明显优于双三次Bezier曲面插值拟合法。然后改进求取K近邻获取拓扑信息的算法,也得到了良好的精简效果。

To obtain the topology structure of local feature is the key point in the point cloud process based on massive measurement,and the efficiency of the algorithm is especially important.For the point cloud that lacked sufficient topology information,the geometry topology information of adjacent points of each data point in the point cloud was first established.Experimental results showed that random Hough alteration pipe reconstruction method established could achieve better performance on the surface fitting of timber trunk compared with the Bezier method.Then the proposed algorithm showed the reasonability when it was used to simplify the processing of K nearest neighbors searching.