中文摘要：

随着大规模点云数据的大量涌现,点云简化问题成为数字几何处理领域的研究热点.本文提出一种基于二次误差度量的自适应点云简化方法.该方法首先提取原始点云的特征点,并对其进行强制保护;而后以非特征点为球心,采用基于二次误差度量的方法,并结合曲率信息计算非特征点覆盖球的半径和最优简化点;最后用最优点代替覆盖球内贡献较少的非特征点,对模型进行自适应简化.该方法不仅具有较快的简化速度,同时还可有效地保持原始点云的几何特征.

英文摘要：

Due to more and more large-scale point-based models emerge in digital geometry processing field,how to simplify the models becomes a challenge for computer graphics scientists.In this paper,we propose an elegant algorithm for decimating point clouds based on quadric error metric.Our algorithm first extracts feature points according to local curvatures of model surfaces,and at the same time tags the feature points to protect them not to be removed during simplication.Then,the algorithm evaluates radii and optimal points of covering balls,which are centered at the non-feature points that are defined based on the quadric error metric and the local curvatures.Finally,our algorithm replaces the non-feature points inside the covering balls by their optimal points and the decimated models are obtained.Experient results show non-feature points that contribute less to defining shape can be simplified in an efficient way,and important geometrical features of original point clouds can be reserved efficiently.