基于FCM-WA联合算法的多种类矿石图像分割

Image Segmentation of Multi Kinds of Ores Based on FCM-WA Joint Algorithm

矿石图像分割是基于机器视觉的矿石粒度分布检测的重要组成部分。针对复合矿山中颜色多样、纹理复杂且边缘粘连的多种类矿石图像难以识别与分割的问题,提出了一种基于FCM-WA联合算法的矿石图像分割方法。首先对矿石图像进行形态学优化,利用双边滤波、直方图均衡化和形态学重构来优化矿石图像的几何特征,减少噪声对分割效果的影响,提高图像对比度;然后将模糊C均值聚类(FCM)算法与分水岭(WA)算法相结合,利用FCM算法进行聚类迭代,计算出合适的分割阈值并对矿石图像进行分割,输出二值化图像;再利用基于距离变换的WA算法优化FCM算法的分割结果,对FCM算法输出的矿石图像边缘粘连部分进行分割,以获取最佳的分割图像。研究结果表明:(1)利用形态学优化流程处理矿石图像能够减少噪声并增强边缘信息,从而提高对比度;(2)相比传统的大津法和遗传算法,本文所提FCM-WA方法的稳健性更强、分割效果更好,对多种类的矿石图像像素分割准确率和矿石粒度识别准确率均可达到92%以上;(3)通过试验验证,FCM-WA方法能够精确地分割颜色多样、纹理特征复杂及边缘粘连的多种类矿石图像,分割结果满足粒度分布检测的要求;(4)FCM-WA方法符合现实矿山企业生产的需求,能够为研发新型矿山智能化粒度检测设备提供可靠的技术支持。

Ore image segmentation is an important part of ore size distribution detection based on machine vision.In order to solve the problem that it is difficult to recognize and segment the multi kinds of ore images with various colors,complex textures and adhesive edges in composite mines,a method of ore image segmentation based on FCM-WA combined algorithm was proposed.Firstly,the ore image is optimized by morphology,which uses bilateral filtering,histogram equalization and morphological reconstruction to optimize the geometric features of the ore image,reduce the impact of noise on the segmentation effect,and improve the image contrast.Then,the FCM algorithm was combined with the watershed algorithm,and the FCM algorithm was used for clustering iteration to calculate the appropriate segmentation threshold,segment the ore image,and output the binary image.Then,the WA algorithm based on distance transformation was used to optimize the segmentation result of FCM algorithm,and the edge conglutination part of ore image output by FCM algorithm was segmented to obtain the best segmentation image.The results show that:(1)Using morphological optimization process to process ore images can reduce noise,enhance edge information and improve contrast.(2)Compared with the traditional Otsu method and genetic algorithm,the FCM-WA method in this paper is more robust and has better segmentation effect.The accuracy of pixel segmentation and ore particle size recognition for multiple kinds of ore images can reach more than 92%.(3)The experiment results show that the FCM-WA method can accurately segment many kinds of ore images with diverse colors,edge adhesion and complex texture features,and the segmentation results meet the requirements of particle size distribution detection.(4)The FCM-WA method in this paper is in line with the production needs of real mining enterprises,and can provide reliable technical support for the development of new mine intelligent particle size detection equipment.