基于模糊聚类算法的孕镶金刚石钻头出刃图像分割

Image segmentation of protruded diamonds on impregnated bits by fuzzy clustering algorithm

本文针对孕镶金刚石钻头唇面彩色图像的特征,采用模糊聚类方法对图像中的色彩进行了聚类分析,以提取出刃的金刚石图像。在图像预处理的基础上,对比了两种图像色彩描述空间对聚类分析的影响,并把模糊聚类分析结果与最大类间方差阈值分割法的分析结果进行了比较。结果表明:模糊聚类方法可对孕镶金刚石唇面的颜色进行有效分类,并准确提取出刃的金刚石图像;用RGB的线性变换量描述金刚石唇面图像,比用Lab色彩空间描述图像,更能得到理想的分割效果;对于孕镶金刚石唇面图像而言,最大类间方差阈值分割法难以区分金刚石颗粒和犁沟,分割效果不如模糊聚类法。

Aiming at the color characteristics of the image of impregnated diamond bit crown, the authors employed fuzzy C-means clustering algorithm to group the colors in the image in order to extract protruded diamond grit images. After preprocessing the image, the effects of two color spaces on clustering were compared, as well as the segmentation effects of fuzzy C-means and Otsu threshold segmentation method. The results show that fuzzy C-means could effectively classify the colors of crown image and correctly extract protruded diamond grit images; better segmentation effect was obtained with linear transformation of RGB expressing crown color than with Lab color space; it＇ s rather difficult to distinguish plow grooves from diamond grits by Otsu threshold segmentation method, so its segmentation effect was not as good as that of fuzzy cltistering algorithm.