基于特征感知的三维点云简化算法研究

3D Point Cloud Simplification Method Based on Feature Perception

点云是表示三维场景的一种方式。与三维网格模型相同,点云在点中包含丰富的信息,如三维空间坐标、颜色、法向量等。但点云中没有保存点与点之间的拓扑关系,也因此没有边与三角形的概念。本文对三维点云场景的简化方法进行研究,针对基于点云数据特点提出特征感知的点云简化方法。根据点云的几何特征和颜色特征总结出基于保留点云局部特征的简化误差度量方法。在实现点云简化中,提出按照点的局部误差大小逐个点迭代并进行点云局部聚类,直到达到简化目标的简化方案。最后实现基于面片的点云可视化。

Point clouds are a way of representing a three-dimensional scene.Like 3d grid models,point clouds contain abundant information in points,such as 3D spatial coordinates,colors,normal vectors,and so on.But the point cloud has no topological relationships between points,and therefore no concepts of sides and triangles.This paper studies the simplification method of 3D point cloud scene and proposes a feature-sensing point cloud simplification method based on the characteristics of point cloud data.According to the geometric and color characteristics of point clouds,a simplified error measurement method based on the local characteristics of reservation point clouds is sum⁃marized.In the simplification of point cloud,a simplified scheme is proposed to iterate one point by one according to the local error of the point and carry out local clustering of point cloud until the simplification goal is achieved.Finally,point cloud visualization based on sur⁃face is realized.