摘要
为了有效地估算点模型的微分属性,提出了一种基于几何特征相似性的估算方法.首先,利用Mean shift(MS)聚类法,对点模型进行几何特征相似性聚类;然后,基于径向基函数(Radial basis functions,RBF),重构各聚类单元的局部隐式曲面;最后,依据经典微分几何理论,在径向基函数曲面上便捷地求解采样点的微分属性并给出具体应用.实验与应用结果表明,该方法能够比较精确地估算出点模型的微分属性且得到有效应用.
Based on the geometry-features similarity,an algorithm is presented for effectively estimating the differential properties on point-sampled surfaces （PSS）. By using mean-shift （MS） clustering,PSS is first clustered into clusters according to the surface-features similarity. Based on radial base functions,a local implicit surface is then reconstructed that approximates the sampled points in a cluster. By applying the theory of classical differential geometry to each implicit surface,the differential properties of each sampled point on PSS are finally estimated and their applications are given. Some experimental results demonstrate that the algorithm can accurately estimate the differential properties on PSS and is effective.
出处
《自动化学报》
EI
CSCD
北大核心
2011年第12期1474-1482,共9页
Acta Automatica Sinica
基金
国家自然科学基金(61073074)
浙江省自然科学基金(Y1090137)
宁波市自然科学基金(2011A610196)资助~~
关键词
微分属性
Mean
shift聚类
径向基函数
点绘制
简化
谷脊线
Differential property
mean-shift clustering
radial basis function （RBF）
point-based rendering
simplifica-tion
valley-ridge line
