PR-MobileOre:基于边界优化与关键点采样的轻量级矿石骨料图像分割算法

刘瑾彧 ,  张高洋 ,  刘安洋 ,  张杨

有色金属(选矿部分) ›› 2026, Vol. 0 ›› Issue (9) : 132 -140.

PDF (1750KB)
有色金属(选矿部分) ›› 2026, Vol. 0 ›› Issue (9) : 132 -140. DOI: 10.20239/j.issn.1671-9492.2026.09.015
矿石特征智能解析专栏

PR-MobileOre:基于边界优化与关键点采样的轻量级矿石骨料图像分割算法

作者信息 +

PR-MobileOre: A Lightweight Segmentation Algorithm for Ore Aggregate Images Based on Boundary Optimization and Keypoint Sampling

Author information +
文章历史 +
PDF (1791K)

摘要

随着智能制造与工业自动化的持续发展,基于图像的矿石骨料粒度检测在选矿过程控制、工程质量监测等领域中正扮演着越来越关键的角色。然而,在实际工业场景中,矿渣干扰、颗粒堆叠、边界模糊等复杂因素广泛存在,严重影响了图像分割方法的准确性与稳定性。此外,受限于设备算力与实时性要求,如何在保障分割精度的同时兼顾模型的计算效率与工业部署适应性,成为亟待解决的关键问题。为此,本文提出一种轻量级、实时性强的矿石图像精细实例分割算法PR-MobileOre。该算法以MobileOne 为骨干网络,融合自主设计的关键点自适应采样机制,依靠不确定性评分聚焦图像边界与模糊区域,并引入边界精细化策略,基于高分辨率特征图完成逐点插值与分类预测,有效提升模型局部结构感知能力与分割边缘一致性。采用多轮迭代优化方式,在维持轻量级模型规模的条件下,实现复杂工业场景下矿石颗粒边界的精准提取。试验结果表明,PR-MobileOre 模型大小仅74. 4 MB,在自建矿石图像数据集上mAP 达到74. 8%,综合精度优于多种主流实例分割算法,兼顾分割高精度与低计算开销,可为矿石骨料智能粒度分析与生产在线控制提供技术支撑。

Abstract

With the continuous advancement of intelligent manufacturing and industrial automation, image-based particle size detection of ore aggregates is playing an increasingly critical role in process control and quality monitoring in mining and construction engineering. However, in real-world industrial scenarios, complex factors such as slag interference, particle stacking, and blurred boundaries are common, severely affecting the accuracy and stability of image segmentation methods. Moreover, due to constraints in computing power and real-time requirements, how to ensure segmentation accuracy while maintaining computational efficiency and deployment adaptability has become a key challenge. To address this, we propose PR-MobileOre, a lightweight and real-time instance segmentation algorithm for fine-grained ore images. This method employs MobileOne as the backbone network and integrates a self-designed keypoint adaptive sampling mechanism, which focuses on image boundaries and ambiguous regions based on uncertainty scoring. It further introduces a boundary refinement strategy to conduct point-wise interpolation and classification prediction on high-resolution feature maps, which effectively improves the capability of local structure perception and edge consistency. With a multi-stage iterative optimization process, the model realizes accurate boundary extraction for ore particles under complex scenarios while keeping a compact model size. Experimental results demonstrate that PR-MobileOre, with only 74. 4 MB of model size, achieves a mAP of 74. 8% on a self-constructed ore image dataset, outperforming various mainstream instance segmentation algorithms. It combines high precision with low computational cost, offering effective technical support for intelligent particle size analysis and real-time control in ore aggregate processing.

关键词

矿石图像分割 / 轻量化网络 / 边界精细化 / 实时图像处理 / 矿石粒度检测

Key words

ore image segmentation / lightweight network / boundary refinement / real-time processing / ore particle size detection

引用本文

引用格式 ▾
刘瑾彧,张高洋,刘安洋,张杨. PR-MobileOre:基于边界优化与关键点采样的轻量级矿石骨料图像分割算法[J]. 有色金属(选矿部分), 2026, 0(9): 132-140 DOI:10.20239/j.issn.1671-9492.2026.09.015

登录浏览全文

4963

注册一个新账户 忘记密码

参考文献

[1]

叶润青, 牛瑞卿, 张良培. 基于多尺度分割的岩石图像矿物特征提取及分析[J]. 吉林大学学报(地球科学版), 2011, 41(4): 1253-1261.

[2]

Ye Runqing, Niu Ruiqing, Zhang Liangpei. Mineral features extraction and analysis based on multiresolution segmentation of petrographic images[J]. Journal of Jilin University (Earth Science Edition), 2011, 41(4): 1253-1261.

[3]

何一东, 陈锐, 吴泽彬, 等. 融合多边形拟合与凹点匹配的黏连重叠矿石图像分割算法[J]. 有色金属(选矿部分), 2024(10): 94-104.

[4]

He Yidong, Chen Rui, Wu Zebin, et al. A segmentation algorithm for adhesive overlapping ore images based on polygon fitting and concave point matching[J]. Nonferrous Metals (Mineral Processing Section), 2024(10): 94-104.

[5]

李文博, 郝兵, 赵虎, 等. 矿石粒度图像在线分析系统的研究与应用[J]. 矿山机械, 2021, 49(4): 47-50.

[6]

Li Wenbo, Hao Bing, Zhao Hu, et al. Research and application of online ore particle size image analysis system[J]. Mining & Processing Equipment, 2021, 49(4): 47-50.

[7]

许文祥, 张国英, 蒋焱, 等. 基于二项分布的双窗OTSU的矿石分割模型[J]. 有色金属(矿山部分), 2019, 71(3): 96-102.

[8]

Xu Wenxiang, Zhang Guoying, Jiang Yan, et al. Ore segmentation model by bi-window OTSU based on binomial distribution[J]. Nonferrous Metals (Mining Section), 2019, 71(3): 96-102.

[9]

危发文. 金属露天矿区传送带破碎矿石粒度检测方法研究及应用[D]. 西安: 西安建筑科技大学, 2022.

[10]

Wei Fawen. Research and application of ore particle size detection method for conveyor belt crushing in metal open-pit mining area[D]. Xi'an: Xi'an University of Architecture and Technology, 2022.

[11]

曾中华, 曹东. 基于改进Unet 的矿石图像分割[J]. 电子测量技术, 2023, 46(21): 176-182.

[12]

Zeng Zhonghua, Cao Dong. Ore image segmentation based on improved Unet[J]. Electronic Measurement Technology, 2023, 46(21): 176-182.

[13]

杨克虎, 龙启航, 汪嘉文, 等. 基于自注意力机制的矿井次光照图像语义分割研究[J]. 矿业安全与环保, 2023, 50(5): 9-18.

[14]

Yang Kehu, Long Qihang, Wang Jiawen, et al. Research on semantic segmentation of mine sub-illumination images based on self-attention mechanism[J]. Mining Safety & Environmental Protection, 2023, 50(5): 9-18.

[15]

刘惠中, 宁剑, 邹起华, 等. 基于改进YOLOv5 的螺旋选矿机矿物分带图像分割算法研究[J]. 有色金属(选矿部分), 2024(1): 96-105.

[16]

Liu Huizhong, Ning Jian, Zou Qihua, et al. Research on a spiral concentrator mineral zoning image segmentation algorithm based on improved YOLOv5[J]. Nonferrous Metals (Mineral Processing Section), 2024(1): 96-105.

[17]

商梦石, 贾瑞强, 张海林. 图像处理技术在矿石粒度检测中的应用[J]. 矿产综合利用, 2016, 37(6): 13-16.

[18]

Shang Mengshi, Jia Ruiqiang, Zhang Hailin. Application of image processing technology in ore particle size measurement[J]. Multipurpose Utilization of Mineral Resources, 2016, 37(6): 13-16.

[19]

Vasu P K A, Gabriel J, Zhu J, et al. MobileOne: an improved one millisecond mobile backbone[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Vancouver, BC, Canada. IEEE, 2023: 7907-7917.

[20]

Rezatofighi H, Tsoi N, Gwak J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Long Beach, CA, USA. IEEE, 2019: 658-666.

[21]

Milletari F, Navab N, Ahmadi S A. V-net: fully convolutional neural networks for volumetric medical image segmentation[C]// 2016 Fourth International Conference on 3D Vision(3DV). Stanford, CA, USA. IEEE, 2016: 565-571.

[22]

He Kaiming, Gkioxari G, Dollar P, et al. Mask R-CNN[C]// 2017 IEEE International Conference on Computer Vision(ICCV). Venice. IEEE, 2017: 2980-2988.

[23]

Bolya D, Zhou Chong, Xiao Fanyi, et al. YOLACT: real-time instance segmentation[C]// 2019 IEEE/CVF International Conference on Computer Vision(ICCV). Seoul, Korea. IEEE, 2019: 9156-9165.

[24]

Wang Xinlong, Zhang Rufeng, Kong Tao, et al. SOLOv2: dynamic and fast instance segmentation[C]// Advances in Neural Information Processing Systems. Red Hook: Curran Associates, 2020, 33: 17721-17732.

[25]

Fang Yuxin, Yang Shusheng, Wang Xinggang, et al. Instances as queries[C]// 2021 IEEE/CVF International Conference on Computer Vision(ICCV). Montreal, QC, Canada. IEEE, 2021: 6890-6899.

[26]

Cheng Tianheng, Wang Xinggang, Chen Shaoyu, et al. Sparse instance activation for real-time instance segmentation[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). New Orleans, LA, USA. IEEE, 2022: 4423-4432.

[27]

Tian Zhi, Shen Chunhua, Chen Hao. Conditional convolutions for instance segmentation[C]// Computer Vision-ECCV 2020. Cham: Springer, 2020: 282-298.

[28]

Tian Zhi, Shen Chunhua, Wang Xinlong, et al. BoxInst: high-performance instance segmentation with box annotations[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Nashville, TN, USA. IEEE, 2021: 5439-5448.

[29]

Lyu Chengqi, Zhang Wenwei, Huang Haian, et al. RTMDet: an empirical study of designing real-time object detectors[EB/OL].(2022-12-16)[2026-08-26]. https://doi.org/10.48550/arXiv.2212 07784.

[30]

Cheng Bowen, Misra I, Schwing A G, et al. Masked-attention mask transformer for universal image segmentation[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). New Orleans, LA, USA. IEEE, 2022: 1280-1289.

基金资助

矿冶过程智能优化制造全国重点试验室和矿冶过程自动控制技术北京市重点实验室开放基金课题(BGRIMM-KZSKL-2023-11)

湖北省自然科学基金资助项目(2024AFB153)

AI Summary AI Mindmap
PDF (1750KB)

0

访问

0

被引

详细

导航
相关文章

AI思维导图

/

〈 〉