基于预处理的图像分割算法在矿岩颗粒上的应用研究  

Research on Application of Image Segmentation Algorithm Based on Pre-Processing in Rock

作　　者：杨时建 周曼 刘得斌 刘志勇 Yang Shijian;Zhou Man;Liu Debin;Liu Zhiyong(SUPCON Co.,Ltd.,Hangzhou 310053)

机构地区：[1]中控技术股份有限公司,杭州市310053

出　　处：《仪器仪表标准化与计量》2023年第6期18-20,共3页Instrument Standardization & Metrology

摘　　要：矿岩颗粒因大小不规则,且表面经常多灰,导致自动识别的难度大、准确率低,而现行的人工识别的方法效率低下。本文针对工业现场的矿石颗粒大小检测,通过对原始图像进行腐蚀膨胀以及中值滤波、高斯滤波和同态滤波等多种滤波方法对比实验后获取较好的预处理效果,再将预处理后的图像采用Canny算法边缘分割、形态学骨架法分割以及分水岭算法进行图像分割对比实验,并将获取的分割图计算轮廓的面积和周长的平方比不超过设定阈值,并对满足该条件的轮廓求最小外接圆,然后找出画面内半径最大的外接圆。Because of the irregular size of ore and rock particles and the often dusty surface,the automatic identification is difficult and the accuracy is low,while the current manual identification method is inefficient.In the paper,for the detection of rock diameter size in the industrial field,through the comparison experiment of many filtering methods such as corrosion expansion,median filtering,Gaussian filtering and homomorphic filtering on the original image,better pre-processing effect is obtained.Then,the edge segmentation of Canny algorithm,morphological skeleton segmentation and watershed algorithm are used for the image segmentation comparison experiment.The ratio of the area and square perimeter of the calculated contour of the obtained segmentation image does not exceed the set threshold,and the minimum circumscribed circle is calculated for the contour that meets this condition,and then the circumscribed circle with the largest radius in the image is found.

关 键 词：自适应二值化 滤波 CANNY算法 形态学骨架法 分水岭算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TF041[自动化与计算机技术—计算机科学与技术]