基于自适应带宽均值漂移聚类算法设计传递函数

Designing Transfer Function Based on Adaptive Bandwidth Mean Shift Clustering Algorithm

摘要: 本文采用均值漂移聚类算法为三维数据场设计传递函数．首先根据采样点的分布提出了两种自适应的带宽计算方法；然后利用梯度幅值对采样点加权，实现数据场的物质分类；最后对分类的结果设计高斯型的映射函数．实验表明，与固定带宽均值漂移算法相比，该方法提高了算法的速度和分类的准确性，并获得了高质量的绘制效果．

Abstract: Based on the mean shift clustering algorithm,the transfer function for three dimensional data set is designed.Firstly two adaptive bandwith analysis methods are proposed according to the distribution of data set,and then the material classification is implemented by including the edge gradient magnitude weighted samples.Finally,a Gaussian transfer function is designed for high-quality rendering.The experiments prove that,compared with the fixed bandwidth mean shift clustering algorithm,the proposed method can get better classification results and higher quality volume rendering.