结合Nystrom方法的三维网格模型分割方法

3D mesh segmentation method combined with Nyström method

针对谱聚类三维网格模型分割方法耗时长、占用内存大的问题,本文提出了一种结合Nystrom方法的三维网格模型分割方法。首先,对模型面心进行采样,计算采样点和所有面心的亲和力数值,使用Nystrom方法估计亲和力矩阵的主特征向量,避免了计算亲和力矩阵的巨大开销;其次,使用K-Means算法对主特征向量聚类,实现对模型的分割;最后,使用自适应邻域滤波算法对分割结果进行优化,去除估计误差。在细分后的普林斯顿数据集上进行实验,并同5种分割方法进行定量比较,结果表明本文方法可以有效降低谱聚类方法的时间、空间开销,并且兰德分数比其余方法平均高0.21,可以得到更高精度的分割结果。

Aiming at the massive time and space overhead of the spectral clustering mesh segmentation method,a mesh segmentation method combined with the Nyström method is proposed.Firstly,sample the model face centers,then calculate the affinity values between the sampled points and all face centers.Using the Nyström method to estimate the principal eigenvectors of the affinity matrix avoids the massive overhead of computing the affinity matrix.Secondly,use the K-Means algorithm to cluster the principal eigenvectors,implementing the segmentation of the model.Finally,the segmentation results are optimized using the adaptive neighborhood filtering algorithm to remove the estimation errors.Experiments on the subdivided Princeton dataset are conducted,and five general segmentation methods are compared.The results show that the proposed method can effectively reduce the time and space overhead of the spectral clustering method.The Rand score is 0.21 higher than the other methods on average,which shows that the proposed method can get more meaningful segmentation results.