基于K-means的金属矿物光片嵌布粒度测量

目前金属矿物嵌布粒度的测定主要依靠技术人员在显微镜下观测,存在工作量大、观测粒度准确度和精度不高等问题。因此提出了基于K-means聚类算法的金属矿物嵌布粒度测量方法。K-means算法初始聚类中心数K值的准确预设是本方法的核心之一,本项目对此开展了深入研究,发现利用金属矿物光片显微镜下图像中的颜色特征,对原矿图像中的矿物种类进行准确判别,再将识别到的矿物颜色特征作为设置K值的依据。通过K-means算法将目的矿物的颜色聚类为单一的颜色,对聚类的图像进行颜色分割,最后利用Feret Diameter精确测量出该金属矿物嵌布粒度。结果表明：该方法明显提高了金属矿物嵌布粒度测量的准确度、精度和效率,具有极大的应用价值。

Particle size measurement of metallic minerals polished section based on K-means

At present, the determination of the particle size of metal minerals inlays mainly relies on the observation of technicians under a microscope, which has problems such as heavy workload and low accuracy and precision of the observed particle size. Therefore, a particle size measurement method based on K-means clustering algorithm is proposed. The accurate presetting of the K value of the initial clustering center number of the K-means algorithm is one of the cores of this method. This project has carried out in-depth research on this and found that the color features in the image under the light sheet microscope of metal minerals can be used for the run of mine image. Accurately distinguish the types of minerals, and then use the identified mineral color characteristics as the basis for setting the K value. The color of the target mineral is clustered into a single color through the K-means algorithm, the clustered image is color segmented, and the Feret Diameter is used to accurately measure the particle size of the metal mineral patch. The results show that this method obviously improves the accuracy, precision and efficiency of the particle size measurement of metallic mineral intercalation, and has great application value.