基于岩块轮廓属性的爆堆图像自适应分割方法  

Self-adaptive blasting rock image segmentation method based on rock block contour attribute

作　　者：郭钦鹏 相志斌 杨仕教 王昱琛 尹裕 GUO Qin-peng;XIANG Zhi-bin;YANG Shi-jiao;WANG Yu-chen;YIN Yu(School of Resources Environment and Safety Engineering,University of South China,Hengyang 421000,Hunan,China;China Nonferrous Metal Changsha Survey and Design Institute Co.,LTD.,Changsha 410000,China)

机构地区：[1]南华大学资源环境与安全工程学院,湖南衡阳421000 [2]中国有色金属长沙勘察设计研究院有限公司,长沙410000

出　　处：《工程爆破》2023年第5期64-71,共8页Engineering Blasting

基　　金：湖南省研究生科研创新基金资助项目(QL20210216,CX20200916)。

摘　　要：针对爆堆岩块图像中因粘连、堆叠、边缘模糊等造成的错误分割问题,提出基于岩块轮廓属性的爆堆图像自适应分割方法。首先对爆堆图像进行预处理,然后采用Phansalkar方法进行二值分割,并采用形态学优化和面积滤波去除噪点,再利用爆堆岩块的轮廓坚实度和迭代腐蚀相结合的方法来标记种子点,最后基于标记的种子点利用分水岭算法对图像进行分割。将该方法用于爆堆图像分割,种子点标记结果表明基于岩块轮廓坚实度的种子点标记方法可避免部分噪点的影响,提高对爆堆岩块标记效率。分割结果表明该方法获得的面积累计曲线与人工分割的面积累计曲线高度相似,3个特征面积参数的最大相对误差仅为4.32%,对于100 cm 2以上的岩块,分割准确率为98.33%。相较于其他用于岩块分割的分水岭改进方法有效地减小了错误分割的可能,实现了基于岩块灰度特征和轮廓特征的爆堆图像高精度自适应分割。To address the problem of incorrect segmentation caused by adhesion,stacking and blurred edges in blasting rock images,we propose an adaptive segmentation method for blasting rock images based on rock block contour properties.Firstly,the method preprocesses the blasting rock images,then uses the Phansalkar method for binary segmentation of images,and uses morphological optimization and area filtering to remove noise.Secondly,uses a combination of the solidity of rock block contour and iterative erosion to mark the seed points.And finally,the image is segmented based on the marked seed points using the watershed algorithm.The method is applied to the blasting rock image segmentation,and results show that the seed point marking method based on the solidity of the rock block contour can avoid the influence of some noise and improve the efficiency of labeling the blasted rock.The segmentation results show that the area accumulation curve obtained by this method is highly similar to that of manual segmentation,and the maximum relative error of the three characteristic area parameters is only 4.32%,and the segmentation accuracy is 98.33%for rock blocks above 100 cm 2.Compared with other watershed improvement methods for rock block segmentation,it effectively reduces the possibility of incorrect segmentation,achieving a high-precision adaptive segmentation of the blasting rock images based on the grayscale features and contour features of the rock blocks.

关 键 词：爆堆图像 Phansalkar 岩块轮廓 种子点标记 分水岭算法 

分 类 号：TD235[矿业工程—矿井建设]