基于多阈值的工业CT体数据自动分类方法

工业CT体数据分类的合理性直接影响到三维体重构的效果。目前的多数分类方法都是交互式的半自动法。文中提出一种实现体数据自动分类的方法,该方法在增强峰谷特性后的体数据灰度分布图上计算差分曲线的过零点,初步判定体数据的类别,再引入独立峰判断准则和缓坡再划分准则,通过类别合并和类别细分确定多个阈值,从而划分体数据类别。该方法能快速、合理地对工业CT体数据进行自动分类。

Automatic Classification Method for Volume Data from Industrial CT Based on Multi-threshold

Classification for volume data from industrial computed tomography affects the result of volume rendering. Most of the methods for classification are semi-automatic ones based on interaction.This paper presents an automatic classification method to class the volume data.In this work,the histogram function of the volume data is built and the volume data is smoothed to extend its kurtosis,then the zero on the differential curve is computed based on the histogram and the classes of the volume data is obtained cursorily,the final classification for volume data is sorted automatically on the rules of single-peak and ramp-repartition.This method can automatically,fleetly and availably cl assify the volume data from industrial CT.