铜锡多金属矿矿石矿物显微图像智能识别与应用

张聿隆 ,  肖仪武 ,  张钰 ,  王名越 ,  权鸿雁

有色金属(选矿部分) ›› 2026, Vol. 0 ›› Issue (9) : 113 -121.

PDF (1711KB)
有色金属(选矿部分) ›› 2026, Vol. 0 ›› Issue (9) : 113 -121. DOI: 10.20239/j.issn.1671-9492.2026.09.013
矿石特征智能解析专栏

铜锡多金属矿矿石矿物显微图像智能识别与应用

作者信息 +

Intelligent Recognition and Application of Microscopic Images of Copper-Tin Polymetallic Ore Minerals

Author information +
文章历史 +
PDF (1751K)

摘要

随着人工智能技术快速发展,基于图像处理的矿物显微图像智能识别,可自动获取光学显微镜下矿石矿物嵌布粒度、解离度等工艺矿物学关键参数,能够减轻人工镜下统计工作量,降低人为带来的主观误差,提升测试数据的客观性与分析效率。某铜锡多金属矿以铜、锡为主要回收元素,伴生锌;主要金属矿物为锡石、黄铜矿、闪锌矿、黄铁矿,含少量蓝辉铜矿、铜蓝等次生硫化铜矿物,非金属矿物主要为石英、绿泥石。本文利用矿冶科技集团开发的矿物特征参数分析系统(MCPAS),融合HSV色彩阈值分割与深度学习注意力机制算法,对矿石原样以及-0.074 mm占50%~80%四组不同磨矿细度产品光片开展图像采集、矿物识别与定量统计,测定主要金属矿物嵌布粒度组成、单体解离、连生及包裹特征。测试结果表明,矿石锡石嵌布粒度粗细分布不均,铜矿物集合体粒度整体偏粗,闪锌矿粒度偏细;锡石、蓝辉铜矿、铜蓝性脆易碎,磨矿过程易发生过磨。结合矿石工艺矿物学特征,建议在-0.074 mm占60%的磨矿细度条件下优先分选已单体解离铜矿物,得到高品质铜精矿;选铜后将硫化物矿物富集再磨分选,产出另一部分铜精矿、锌精矿和硫精矿;浮选尾矿采用重选回收锡石,细粒锡石辅以浮选回收。

Abstract

With the rapid development of artificial intelligence technology, intelligent recognition of mineral micro-images based on image processing can automatically obtain key process-mineralogical parameters such as grain-size distribution and mineral liberation degree under an optical microscope. It can reduce the workload of manual microscopic statistics, lower subjective human errors, and improve the objectivity and analysis efficiency of test data. For a copper-tin polymetallic ore, copper and tin are the main recovery elements with associated zinc. Its main metallic minerals are cassiterite, chalcopyrite, sphalerite and pyrite, together with a small amount of secondary copper sulfide minerals including digenite and covellite, and its main non-metallic minerals are quartz and chlorite. In this paper, the Mineral Characteristic Parameter Analysis System (MCPAS) developed by BGRIMM was adopted. Combined with HSV color threshold segmentation and deep-learning attention-mechanism algorithm, image acquisition, mineral identification and quantitative statistics were carried out on raw ore and four groups of grinding products with grinding fineness of -0.074 mm accounting for 50%-80%. The grain-size composition, monomer liberation, intergrowth and inclusion characteristics of major metallic minerals were determined. The test results show that cassiterite has uneven grain-size distribution, copper mineral aggregates are relatively coarse-grained, while sphalerite is fine-grained. Cassiterite, digenite and covellite are brittle and friable, and are prone to over-grinding. According to the process-mineralogical characteristics of the ore, it is recommended to preferentially separate monomer-dissociated copper minerals at the grinding fineness of -0.074 mm accounting for 60% to obtain high-quality copper concentrate. After copper flotation, sulfide minerals are enriched for regrinding and separation to produce another copper concentrate, zinc concentrate and sulfur concentrate. Cassiterite is recovered by gravity separation from flotation tailings, and fine-grained cassiterite is further recovered by flotation.

关键词

铜锡多金属矿 / 矿物图像 / 智能识别 / 工艺矿物学 / 自动分析

Key words

copper-tin polymetallic ore / mineral image / intelligent recognition / process mineralogy / automatic analysis

引用本文

引用格式 ▾
张聿隆,肖仪武,张钰,王名越,权鸿雁. 铜锡多金属矿矿石矿物显微图像智能识别与应用[J]. 有色金属(选矿部分), 2026, 0(9): 113-121 DOI:10.20239/j.issn.1671-9492.2026.09.013

登录浏览全文

4963

注册一个新账户 忘记密码

参考文献

[1]

肖仪武. 选矿工艺矿物学[M]. 北京: 科学出版社, 2020.

[2]

Xiao Yiwu. Process mineralogy[M]. Beijing: Science Press, 2020.

[3]

周乐光. 工艺矿物学[M].2版. 北京: 冶金工业出版社, 2002.

[4]

Zhou Leguang. Process mineralogy[M]. 2nd ed. Beijing: Metallurgical Industry Press, 2002.

[5]

肖仪武. 中国选矿工艺矿物学发展历程、研究现状与展望[J]. 有色金属(选矿部分), 2019(5): 6-8.

[6]

Xiao Yiwu. Development history, research status and outlook of process mineralogy in China[J]. Nonferrous Metals (Mineral Processing Section), 2019(5): 6-8.

[7]

潘兆橹. 结晶学及矿物学: 下册[M].3版. 北京: 地质出版社, 1994.

[8]

Pan Zhaolu. Crystallography and mineralogy: Volume Ⅱ[M]. 3rd ed. Beijing: Geological Publishing House, 1994.

[9]

肖仪武, 叶小璐, 冯凯, 等. “四稀”金属矿工艺矿物学研究技术现状与展望[J]. 矿冶, 2022, 31(3): 6-13.

[10]

Xiao Yiwu, Ye Xiaolu, Feng Kai, et al. Current situation and prospect of “Four Rare” process mineralogy research technology[J]. Mining and Metallurgy, 2022, 31(3): 6-13.

[11]

肖仪武, 方明山, 付强, 等. 工艺矿物学研究的新技术与新理念[J]. 矿产保护与利用, 2018, 38(3): 49-54.

[12]

Xiao Yiwu, Fang Mingshan, Fu Qiang, et al. New techniques and concepts in process mineralogy[J]. Conservation and Utilization of Mineral Resources, 2018, 38(3): 49-54.

[13]

周开军, 桂卫华, 朱红求. 矿物浮选泡沫图像分割参数的最优性分析[J]. 华中科技大学学报(自然科学版), 2013, 41(增刊1): 432-435.

[14]

Zhou Kaijun, Gui Weihua, Zhu Hongqiu. Parameter optimality analysis for mineral floatation froth image segmentation[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2013, 41(S1): 432-435.

[15]

牛福生, 张悦, 聂轶苗. 图像处理技术在工艺矿物学研究中的应用[J]. 金属矿山, 2010, 39(5): 92-95, 103.

[16]

Niu Fusheng, Zhang Yue, Nie Yimiao. Applications of image processing techniques in process mineralogy[J]. Metal Mine, 2010, 39(5): 92-95, 103.

[17]

Dias N M P, Gonçalves D, Leite W C, et al. Morphological characterization of soil clay fraction in nanometric scale[J]. Powder Technology, 2013, 241: 36-42.

[18]

Kim S, Ree Jinhan. Quantification of the spatial distribution of mineral phases and grains in rock using a 2-D multiple-area density map technique[J]. Tectonophysics, 2012, 522/523: 176-186.

[19]

Kistner M, Jemwa G T, Aldrich C. Monitoring of mineral processing systems by using textural image analysis[J]. Minerals Engineering, 2013, 52: 169-177.

[20]

李洪艳, 曹建荣, 谈文婷, 等. 图像分割技术综述[J]. 山东建筑大学学报, 2010, 25(1): 85-89, 93.

[21]

Li Hongyan, Cao Jianrong, Tan Wenting, et al. Review of image segmentation technology[J]. Journal of Shandong Jianzhu University, 2010, 25(1): 85-89, 93.

[22]

徐丽燕, 刘复昌, 曹国, 等. 基于边缘流与区域归并的彩色图像分割方法[J]. 光电子·激光, 2011, 22(10): 1582-1587.

[23]

Xu Liyan, Liu Fuchang, Cao Guo, et al. Color image segmentation algorithm based on edgeflow and region merging[J]. Journal of Optoelectronics · Laser, 2011, 22(10): 1582-1587.

[24]

张建明, 张菊, 王娟. 基于梯度修正和区域融合的分水岭分割算法[J]. 计算机应用, 2011, 31(2): 369-371.

[25]

Zhang Jianming, Zhang Ju, Wang Juan. Watershed segmentation algorithm based on gradient modification and region merging[J]. Journal of Computer Applications, 2011, 31(2): 369-371.

[26]

陈丽芳. 融合改进分水岭和区域生长的彩色图像分割方法[J]. 计算机工程与科学, 2013, 35(4): 93-98.

[27]

Chen Lifang. Color image segmentation by combining improved watershed and region growing[J]. Computer Engineering & Science, 2013, 35(4): 93-98.

[28]

黄永进, 何剑锋, 李卫东, 等. 基于改进Swin-Transformer模型的铜矿X射线图像分类研究[J]. 有色金属(选矿部分), 2024(6): 116-124.

[29]

Huang Yongjin, He Jianfeng, Li Weidong, et al. Research on copper ore X-ray image classification based on improved Swin-Transformer model[J]. Nonferrous Metals (Mineral Processing Section), 2024(6): 116-124.

[30]

张臻悦, 蒋灵, 郭文达, 等. 智能技术在矿物加工领域中的应用进展[J]. 有色金属(选矿部分), 2023(6): 1-8, 18.

[31]

Zhang Zhenyue, Jiang Ling, Guo Wenda, et al. Application progress of intelligent technology in mineral processing field[J]. Nonferrous Metals (Mineral Processing Section), 2023(6): 1-8, 18.

[32]

司佳, 陈思平, 袁洲, 等. 基于图像识别与生成技术的人工智能技术应用[J]. 科技资讯, 2023, 21(22): 47-50.

[33]

Si Jia, Chen Siping, Yuan Zhou, et al. Application of artificial intelligence technology based on image recognition and generation technology[J]. Science & Technology Information, 2023, 21(22): 47-50.

[34]

胡雅祺, 孔静, 李宇恒, 等. 基于深度学习的图像识别技术在选矿中的应用进展[J]. 矿冶, 2024, 33(1): 131-140.

[35]

Hu Yaqi, Kong Jing, Li Yuheng, et al. Progress on the application of image recognition technology based on deep learning in mineral processing[J]. Mining and Metallurgy, 2024, 33(1): 131-140.

[36]

付文博, 孙涛, 梁藉, 等. 深度学习原理及应用综述[J]. 计算机科学, 2018, 45(增刊1): 11-15, 40.

[37]

Fu Wenbo, Sun Tao, Liang Ji, et al. Review of principle and application of deep learning[J]. Computer Science, 2018, 45(S1): 11-15, 40.

[38]

郑远攀, 李广阳, 李晔. 深度学习在图像识别中的应用研究综述[J]. 计算机工程与应用, 2019, 55(12): 20-36.

[39]

Zheng Yuanpan, Li Guangyang, Li Ye. Survey of application of deep learning in image recognition[J]. Computer Engineering and Applications, 2019, 55(12): 20-36.

[40]

邱俊航. 基于人工智能算法的图像识别技术分析[J]. 数字通信世界, 2024(12): 25-27.

[41]

Qiu Junhang. Research on graphics and image processing technology based on artificial intelligence[J]. Digital Communication World, 2024(12): 25-27.

[42]

孙传尧, 周俊武, 贾木欣, 等. 基因矿物加工工程研究[J]. 有色金属(选矿部分), 2018(1): 1-7.

[43]

Sun Chuanyao, Zhou Junwu, Jia Muxin, et al. Research on genetic mineral processing engineering[J]. Nonferrous Metals (Mineral Processing Section), 2018(1): 1-7.

[44]

Liu Ze, Lin Yutong, Cao Yue, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]∥ 2021 IEEE/CVF International Conference on Computer Vision (ICCV). Montreal, Canada. IEEE, 2021: 9992-10002.

[45]

Song Wei, Zheng Nan, Liu Xiangchun, et al. An improved U-Net convolutional networks for seabed mineral image segmentation[J]. IEEE Access, 2019, 7: 82744-82752.

[46]

Zhang Wei, Borji A, Wang Zhou, et al. The application of visual saliency models in objective image quality assessment: a statistical evaluation[J]. IEEE Transactions on Neural Networks and Learning Systems, 2016, 27(6): 1266-1278.

[47]

Ronneberger O, Fischer P, Brox T. U-Net: convolutional networks for biomedical image segmentation[C]∥ Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015. Cham: Springer, 2015: 234-241.

[48]

Badrinarayanan V, Kendall A, Cipolla R. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495.

[49]

Zhao Hengshuang, Shi Jianping, Qi Xiaojuan, et al. Pyramid scene parsing network[PP/OL]. V2. arXiv(2017-04-27)[2025-09-29]. https://doi.org/10.48550/arXiv.1612.01105.

基金资助

云南省科技计划项目(202504BW050001)

AI Summary AI Mindmap
PDF (1711KB)

0

访问

0

被引

详细

导航
相关文章

AI思维导图

/

〈 〉