基于K均值聚类的体绘制传递函数设计方法研究

RESEARCH ON VOLUME RENDERING TRANSFER FUNCTION DESIGN BASED ON K-MEANS CLUSTERING

提出一种基于K均值聚类算法的体绘制多维传递函数设计方法。在利用灰度—梯度直方图分析体数据内部结构信息的基础上,应用K均值聚类算法对整个体数据进行聚类分类,对属于不同聚类中的体素值进行伪彩色映射,实现体数据与彩色编码的转换关系。实验表明,利用该方法所设计的体绘制传递函数能够揭示体数据的内部结构关系,重建的三维图像逼真、质量高。

This paper proposes a design method of multi-dimensional transfer function of volume rendering based on K-means clustering algorithm.On the basis of analysing the internal structural information of volume data with gray-gradient histogram,the whole volume data was clustered and classified using K-means clustering algorithm.For the volume data belonging to different clustering,the pseudo-colour mapping was conducted to achieve the transformation between volume data and colour codes.Experimental results showed that the volume rendering transfer function designed by the proposed method can reveal the internal structure relations of volume data,and the reconstructed three-dimensional images were fidelity with high quality.