一种基于模糊聚类的海量测量数据简化方法

Data Simplification of Point Sampled Geometry Based on Fuzzy Clustering

为满足模型简化的几何与工程信息保真性要求,提出一种基于模糊聚类分析原理、直接从海量散乱点简化的方法。通过引入对采样点的几何相似性隶属度来表征被测曲面形状的自然变化,使简化点集倾向于聚集在陡峭区域,并以强制约束相似性隶属度反映设计者对简化点集的工程要求,利于保留点集中蕴含的工程细节特征和指导曲面的后续处理。实验表明,简化导致的形状细节和工程细节损失可得到有效抑制。

To meet the demand for both geometric and engineering fidelity of the simplification, a new simplification method, which simplifies data directly form dense scattered data points on the basis of fuzzy clustering without intermediate tessellation, is presented. By introducing the description of a samples fuzzy geometric attribute, the ambiguity of segmentation and consolidation of samples can be expressed. The reduced points are inclined to gather at regions of high curvature and surface boundaries. By introducing fuzzy imperative constraint attribute, engineering requests of the designer can be considered and satisfied. Detail features with important machining values of the measured object can be preserved. And post-processing the measured object can be better instructed. Experiments show that the loss of shape details and engineering information can be well controlled during data simplification.