摘要
用选矿机进行矿物分选可以提高目标矿物的含量,从而优化资源利用,提高生产效率,在矿业领域中具有重要意义与价值,但分选过程中通过X射线透射技术所得的矿石图像会存在目标黏连及重叠的情况,这将严重影响矿物分选效率及精度,获取独立矿石图像是使用该方法对矿石进行识别、定位、分选的必要条件。为提高矿物分选精度与效率,提出了一种基于多边形拟合的凹点检测与匹配算法用于分割黏连矿石图像。首先,将矿石二值图像拟合为多边形图像并从图像拐点中甄别出凹点;其次,利用拐点构建直线方程形成检测区域,在该区域内搜寻凹点的待连接点,完成凹点匹配;最后,将计算机视域由二值图转移至灰度图,通过分析待分割区域的灰度值判断该区域是否为矿石间实际黏连处,从而决定是否执行分割操作。试验结果表明,算法在黏连矿石数据集上的总体分割准确率为93.60%,凹点噪声率仅为5.23%,处理尺寸为281×336的图像平均计算时间低至5.13 ms,检测效率最高可提升220倍以上,其噪声凹点滤除能力、凹点检测精度、分割准确率、算法运行速率均优于同类算法,对不同黏连数量及不同黏附形式的矿石图像均展现出较强的分割稳定性。
The use of a beneficiation machine for mineral sorting can increase the content of target minerals,optimize resource utilization,and improve production efficiency,which is of great significance and value in the mining field.However,during the sorting process,the ore images obtained through X-ray transmission technology may have target adhesion and overlap,which will seriously affect the efficiency and accuracy of mineral sorting.Obtaining an independent ore image is a necessary condition for using the method to identify,locate and sort the ore.To improve the accuracy and efficiency of mineral sorting,a concave point detection and matching algorithm based on polygon fitting is proposed for segmenting adhesive ore images.Firstly,fitting a binary image of the ore into a polygon image and identifying concave points from the turning point of the image.Secondly,constructing a linear equation by using a turning point to form a detection area,searching the to-be-connected points of the concave points in the area,and completing concave point matching.Finally,the computer view is transferred from the binary image to the gray image,and the gray value of the region to be segmented is analyzed to determine whether the region is the actual adhesion between ores,so as to determine whether to execute the segmentation operation.The results show that the overall segmentation accuracy of the algorithm on the adhesive ore dataset is 93.60%,the concave noise rate is only 5.23%,and the average calculation time for processing images with a size of 281×336 is as low as 5.13 ms.Its noise concave filtering ability,concave detection accuracy,segmentation accuracy,and algorithm running speed are all better than those of similar algorithms.It shows strong segmentation stability for ore images with different adhesive quantities and forms.
作者
何一东
陈锐
吴泽彬
钟崇贵
王静
HE Yidong;CHEN Rui;WU Zebin;ZHONG Chonggui;WANG Jing(School of Mechanical and Electronic Engineering,East China University of Technology,Nanchang 330013,China;School of Education Science,Xinjiang Normal University,Urumqi 830017,China)
出处
《有色金属（选矿部分）》
CAS
2024年第10期94-104,132,共12页
Nonferrous Metals（Mineral Processing Section）
基金
国家自然科学基金资助项目(12365026)
江西省重点研发计划项目(20232BBE50013)
抚州市揭榜挂帅项目(XMBH00016)。
关键词
矿石图像分割
X射线透射
多边形拟合
凹点检测与匹配
区域搜寻
预分割检测
ore image segmentation
X-ray transmission
polygon fitting
concave point detection and matching
region search
pre segmentation detection
