摘要
矿石图像分割效果受光照条件、目标密集性及边缘对比度低等因素制约,致使大块率统计精度偏低.为此以鞍千矿爆破矿石图像为数据源,首先利用双边滤波算法去除特征增强后的图像噪声,然后分别采用自适应阈值算法和整体嵌套边缘检测(holistically-nested edge detection, HED)算法初步分割矿石图像,再利用形态学和去除连通域算法去除因矿石表面纹理形成的分割孔洞,进一步融合两种分割结果,引入基于距离运算的分水岭算法消除矿石图像欠分割现象,最终实现矿石图像的优化分割.研究结果表明,该方法可有效提高爆破矿石图像分割准确性,实现露天矿爆破大块率精确统计,为爆破效果智能评价提供技术支持.
The ore image segmentation effect is limited by factors such as lighting conditions,targets density and low contrast of edges,resulting in low statistical accuracy of the bulk rate.Anqian Mine blasted ore image was used as the data source,firstly the bilateral filtering algorithm was used to remove the feature-enhanced image noise,then the adaptive thresholding algorithm and HED(holistically-nested edge detection)algorithm were used to initially segment the ore image,and then the morphology and connectivity removal algorithm were used to remove the segmentation hole formed by the texture of the ore surface.The results were further fused and a distance-based watershed algorithm was introduced to eliminate the under-segmentation of the ore image,and finally the optimal segmentation of the ore image was achieved.The results show that the proposed method can effectively improve the accuracy of blasted ore image segmentation,achieve accurate statistics of blasting bulk rate in open pit mines,and provide technical support for intelligent evaluation of blasting effects.
作者
毛亚纯
樊硕
曹旺
李时
MAO Ya-chun;FAN Shuo;CAO Wang;LI Shi(School of Resources&Civil Engineering,Northeastern University,Shenyang 110819,China;School of Civil Engineering,Shenyang Urban Construction University,Shenyang 110167,China)
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2023年第5期705-711,共7页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金资助项目(52074064)
国家重点研发计划项目(2016YFC0801602)。
关键词
爆破大块率
矿石图像分割
HED算法
自适应阈值算法
分水岭算法
blasting bulk rate
ore image segmentation
HED(holistically-nested edge detection)algorithm
adaptive thresholding algorithm
watershed algorithm
作者简介
毛亚纯(1966-),男,辽宁本溪人,东北大学教授.
