基于K-means聚类的磁瓦缺陷图像分割方法

为了实现磁瓦图像中缺陷的准确检测，以分割磁瓦端面崩块缺陷为目的提出一种基于K-means聚类的分割方法。磁瓦图像采集的关键技术是光源选用，分析传统的磁瓦图像分割方法——阈值分割，并以迭代选择阈值算法作为对比算法进行介绍；着重剖析K-means算法的基本聚类原理，并引出其算法实现流程。采用两种算法对磁瓦端面图像进行分割。结果表明，基于K-means聚类算法对磁瓦图像进行分割，能够正确分割出磁瓦端面的崩块缺陷。

A Segmentation Method of Magnetic Tile Defect Image Based on K-means Clustering

In order to realize the accurate detection of defects in magnetic tile images， this paper proposed a segmentation method based on K-means clustering for the purpose of segmenting magnetic tile edge surface broken defects. Firstly， the key technology of magnetic tile images acquisition was the selection of light source， and the image acquisition system used in this research was introduced. Secondly， the traditional method of magnetic tile image segmentation - threshold segmentation was analyzed， and the iterative threshold selection algorithm was used as a comparison algorithm for a simple introduction. Finally， the basic clustering principle of K-means algorithm was emphatically analyzed， and the algorithm implementation process was also introduced. By using two algorithms to segment the edge surface images of the magnetic tile， the results show that the segmentation method adopted in this paper can correctly segment the broken defects of the edge surface of the magnetic tile.