Mine remote sensing image denoising method based on improved YOLOv5
-
摘要:
典型露天矿场景的图像呈现多类型复合噪声特征，信噪比较低且具有显著的空间异质性，现有深度学习模型大多直接迁移自然图像去噪架构，忽视了矿山遥感图像特有的噪声分布规律。针对该问题，提出了一种基于改进YOLOv5的矿山遥感图像去噪方法。针对传统YOLOv5在高噪声环境下性能不稳定的问题，引入了多尺度特征融合模块，以增强模型对不同尺寸噪声的识别能力，同时结合残差注意力机制，提升了模型对有用特征的提取能力，增强了去噪效果的鲁棒性。采用自适应噪声估计技术，根据图像不同区域的噪声特性动态调整去噪参数，实现了更为精准的噪声抑制。实验结果表明：改进YOLOv5在峰值信噪比（PSNR）和结构相似性指数（SSIM）上均显著优于其他经典去噪方法，相较原始YOLOv5，PSNR提高2.5 dB，SSIM提高了0.05；改进YOLOv5在所有噪声类型下均表现出色，尤其是在高斯噪声环境中，其PSNR和SSIM分别达32.5 dB和0.95，显著优于其他经典去噪方法。
Abstract:The images of typical open-pit mining scenarios exhibit multi-type composite noise characteristics, with a low signal-to-noise ratio and significant spatial heterogeneity. Most existing deep learning models directly transfer denoising architectures from natural images, ignoring the unique noise distribution patterns of mining remote sensing images. To address the issue, a mine remote sensing image denoising method based on improved YOLOv5 was proposed. Considering the instability of traditional YOLOv5 in high-noise environments, a multi-scale feature fusion module was introduced to enhance the model's ability to recognize noise of different sizes. Additionally, a residual attention mechanism was incorporated to improve the extraction of useful features and enhance the robustness of the denoising effect. An adaptive noise estimation technique was employed to dynamically adjust denoising parameters based on the noise characteristics of different image regions, achieving more precise noise suppression. The experimental results showed that the improved YOLOv5 significantly outperformed other classical denoising methods in terms of peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM). Compared to the original YOLOv5, the PSNR value increased by 2.5 dB, and the SSIM improved by 0.05. The improved YOLOv5 performed well under all noise types, especially in Gaussian noise environments, where its PSNR and SSIM reached 32.5 dB and 0.95, respectively, significantly surpassing other classical denoising methods.
-
-
![]()
图 1 基于改进YOLOv5的矿山遥感图像去噪方法流程
Figure 1. Process of mine remote sensing image denoising method based on improved YOLOv5
![]()
图 7 改进YOLOv5在高噪声环境下的训练损失
Figure 7. Training loss of improved YOLOv5 in high noise environment
表 1 不同算法的去噪性能比较
Table 1 Comparison of denoising performance of different algorithms
算法 PSNR /dB SSIM 运行时间/ s 参数量/106个 中值滤波 25.3 0.78 0.05 — 维纳滤波 26.7 0.81 0.07 — 非局部均值滤波 27.5 0.83 0.15 — 去噪自编码器 28.9 0.86 1.20 1.2 原始YOLOv5 29.4 0.88 0.95 7.0 改进YOLOv5 31.9 0.93 1.10 7.5 
下载: 导出CSV
表 2 不同噪声类型下各算法的PSNR和SSIM
Table 2 PSNR and SSIM of each method under different noise types
噪声类型 算法 PSNR /dB SSIM 高斯噪声 改进YOLOv5 32.5 0.95 原始YOLOv5 29.4 0.88 去噪自编码器 28.9 0.86 非局部均值滤波 27.5 0.83 维纳滤波 26.7 0.81 中值滤波 25.3 0.78 椒盐噪声 改进YOLOv5 31.2 0.93 原始YOLOv5 27.8 0.85 去噪自编码器 26.5 0.80 非局部均值滤波 27.5 0.83 维纳滤波 26.7 0.81 中值滤波 25.3 0.78 斑点噪声 改进YOLOv5 31.0 0.92 原始YOLOv5 28.0 0.89 去噪自编码器 27.0 0.84 非局部均值滤波 27.5 0.83 维纳滤波 26.7 0.81 中值滤波 25.3 0.78
下载: 导出CSV
表 3 消融实验结果
Table 3 Results of ablation experiments
模块配置 PSNR/dB SSIM 运行时间/s 原始YOLOv5 29.4 0.88 0.95 +多尺度特征融合模块 30.2 0.92 1.00 +多尺度特征融合模块+残差注意力机制 30.8 0.95 1.05 +多尺度特征融合模块+残差注意力机制
+自适应噪声估计31.9 0.93 1.10
下载: 导出CSV
-
[1] 张玉茜,刘文荣,孙勇,等. 基于Faster−RCNN与自注意力机制的矿山图像异常检测算法[J]. 金属矿山,2024(7):196-201. ZHANG Yuqian,LIU Wenrong,SUN Yong,et al. Mining image anomaly detection algorithm based on faster-RCNN and self-attention mechanism[J]. Metal Mine,2024(7):196-201.
[2] XIAO Lingzhi,WANG Shengbiao,ZHANG Jun,et al. Poisson2Poisson-Sparse:unsupervised Poisson noise image denoising based on sparse modeling[J]. Signal Processing,2025,230. DOI:10.1016/j.sigpro. 2024.109870.
[3] GUO Xiaoqiang,LIU Xinhua,ZHANG Xu,et al. A novel denoising approach based on improved invertible neural networks for real-time conveyor belt monitoring[J]. IEEE Sensors Journal,23(3):3194-3203.
[4] 傅荣春雪,刘君强,冯潇楠,等. 基于剪枝和去噪的航空发动机故障图像识别与预测[J]. 航空计算技术,2023,53(2):40-44. DOI: 10.3969/j.issn.1671-654X.2023.02.009 FU Rongchunxue,LIU Junqiang,FENG Xiaonan,et al. Recognition and prediction of aero-engine fault image based on pruning and denoising[J]. Aeronautical Computing Technique,2023,53(2):40-44. DOI: 10.3969/j.issn.1671-654X.2023.02.009
[5] 李龙海,郭华锋,于萍,等. 矿山远程监控图像的改进非局部均值滤波算法[J]. 金属矿山,2022(10):165-169. LI Longhai,GUO Huafeng,YU Ping,et al. Improved non-local means filtering algorithm of mine remote monitoring image[J]. Metal Mine,2022(10):165-169.
[6] 赵鑫,冯敬培. 钢丝绳张力监测系统的图像处理功能设计研究[J]. 煤炭科技,2022,43(3):75-78,82. ZHAO Xin,FENG Jingpei. Research on image processing function design of wire rope tension monitoring system[J]. Coal Science & Technology Magazine,2022,43(3):75-78,82.
[7] 孔华永,娄昭远. 基于生成式对抗网络的煤矿图像去噪研究[J]. 煤矿机械,2022,43(7):173-175. KONG Huayong,LOU Zhaoyuan. Research on coal mine image denoising based on generative adversarial network[J]. Coal Mine Machinery,2022,43(7):173-175.
[8] 车守全,李涛,包从望,等. 矿区遥感图像去噪方法研究[J]. 工矿自动化,2022,48(1):113-118,124. CHE Shouquan,LI Tao,BAO Congwang,et al. Research on denoising method of remote sensing image in mining area[J]. Industry and Mine Automation,2022,48(1):113-118,124.
[9] 王小兵. 融合提升小波阈值与多方向边缘检测的矿区遥感图像去噪[J]. 国土资源遥感,2020,32(4):46-52. WANG Xiaobing. Denoising algorithm based on the fusion of lifting wavelet thresholding and multidirectional edge detection of remote sensing image of mining area[J]. Remote Sensing for Land & Resources,2020,32(4):46-52.
[10] TIAN Chunwei,ZHENG Menghua,ZUO Wangmeng,et al. Multi-stage image denoising with the wavelet transform[J]. Pattern Recognition,2023,134. DOI: 10.1016/j.patcog.2022.109050.
[11] ZHANG Jie,WANG Fengxian,ZHANG Huanlong,et al. Compressive sensing spatially adaptive total variation method for high-noise astronomical image denoising[J]. The Visual Computer,2024,40(2):1215-1227. DOI: 10.1007/s00371-023-02842-w
[12] 沈丹萍,赵爽. 基于轻量化YOLOX−S与多阈值分割的矿山遥感图像去噪算法[J]. 金属矿山,2024(9):175-180. SHEN Danping,ZHAO Shuang. Mine remote sensing image denoising algorithm based on lightweight YOLOX-S and multi-threshold eegmentation[J]. Metal Mine,2024(9):175-180.
[13] 宋怀波,雒鹏鑫,王亚男,等. 融合YOLOv5s与Criminisi算法的农业遥感图像去云方法研究[J]. 西北农林科技大学学报(自然科学版),2023,51(3):143-154. SONG Huaibo,LUO Pengxin,WANG Yanan,et al. Fusion of YOLOv5s network and Criminisi algorithm to remove clouds from agricultural remote sensing images[J]. Journal of Northwest A & F University(Natural Science Edition),2023,51(3):143-154.
[14] 樊培利,王建军,艾薇. 基于生成对抗残差学习的矿山远程监控图像去噪算法[J]. 金属矿山,2024(5):286-292. FAN Peili,WANG Jianjun,AI Wei. Mine remote monitoring image denoising algorithm based on generated residuals learning[J]. Metal Mine,2024(5):286-292.
[15] DU Jiazhi,QIAO Xin,YAN Zifei,et al. Flexible image denoising model with multi-layer conditional feature modulation[J]. Pattern Recognition,2024,152. DOI: 10.1016/j.patcog.2024.110372.
[16] 马琳,苏明,兰义湧. 一种基于多尺度特征复用残差网络的矿山图像重建算法[J]. 金属矿山,2023(11):228-233. MA Lin,SU Ming,LAN Yiyong. A mine image reconstruction algorithm based on multi-scale feature multiplexing residual network[J]. Metal Mine,2023(11):228-233.
[17] 李茹,鲁海峰. 基于多尺度卷积神经网络的矿山监控图像识别研究[J]. 唐山学院学报,2023,36(6):5-10,16. LI Ru,LU Haifeng. Research on mine monitoring image recognition based on multi-scale convolutional neural network[J]. Journal of Tangshan University,2023,36(6):5-10,16.
[18] 江结林,史明月,杨海东,等. 多头注意力机制的全频图像去噪算法[J]. 计算机工程与应用,2024,60(16):236-247. DOI: 10.3778/j.issn.1002-8331.2311-0430 JIANG Jielin,SHI Mingyue,YANG Haidong,et al. Omni-frequency image denoising with multi-head attention[J]. Computer Engineering and Applications,2024,60(16):236-247. DOI: 10.3778/j.issn.1002-8331.2311-0430
[19] WU Wencong,LIU Shijie,XIA Yuelong,et al. Dual residual attention network for image denoising[J]. Pattern Recognition,2024,149. DOI: 10.1016/j.patcog.2024.110291.
[20] SHARIF S M A,ALI NAQVI R,BISWAS M. Learning medical image denoising with deep dynamic residual attention network[J]. Mathematics,2020,8(12). DOI: 10.3390/math8122192.
[21] TAO Haowu,GUO Wenhua,HAN Rui,et al. RDASNet:image denoising via a residual dense attention similarity network[J]. Sensors,2023,23(3). DOI: 10.3390/s23031486.
[22] 张宇,窦银科,赵亮亮,等. MF−TLID:一种多特征融合输电线覆冰图像去噪方法[J]. 重庆理工大学学报(自然科学),2024,38(10):147-155. ZHANG Yu,DOU Yinke,ZHAO Liangliang,et al. An ice-covered transmission line image denoising method fused with multiple features[J]. Journal of Chongqing University of Technology (Natural Science),2024,38(10):147-155.
[23] 余卓璞,周冬明,周联敏. 基于高阶交互的渐进式真实图像去噪网络[J]. 科学技术与工程,2024,24(21):9043-9052. DOI: 10.12404/j.issn.1671-1815.2306226 YU Zhuopu,ZHOU Dongming,ZHOU Lianmin. Progressive real-word image denoising network based on higher-order interactions[J]. Science Technology and Engineering,2024,24(21):9043-9052. DOI: 10.12404/j.issn.1671-1815.2306226
[24] 张杰,卢淼鑫,李嘉康,等. 基于残差密集卷积自编码的高噪声图像去噪方法[J]. 计算机科学,2024,51(增刊1):567-573. ZHANG Jie,LU Miaoxin,LI Jiakang,et al. High-noise image denoising method based on residual dense convolutional autoencoder[J]. Computer Science,2024,51(S1):567-573.
[25] 伏锦,黄山. 基于改进循环生成对抗网络的图像去噪研究[J]. 计算机工程与应用,2023,59(17):178-186. DOI: 10.3778/j.issn.1002-8331.2209-0050 FU Jin,HUANG Shan. Research on image denoising based on improved CycleGAN[J]. Computer Engineering and Applications,2023,59(17):178-186. DOI: 10.3778/j.issn.1002-8331.2209-0050
-
期刊类型引用(27)
1. 翟小伟，李心田，侯钦元，马腾，宋波波，郝乐. 漏风条件下采空区煤自燃分级预警方法研究及应用. 煤炭科学技术. 2025(01): 161-169 . 
百度学术
2. 刘东洋，张浪，姚海飞，徐长富，赵尤信，张逸斌，段思恭. 煤矿火灾智能预警系统研发与应用. 工矿自动化. 2024(01): 1-8+16 . 本站查看
3. 邓军，李鑫，王凯，王伟峰，闫军，汤宗情，康付如，任帅京. 矿井火灾智能监测预警技术近20年研究进展及展望. 煤炭科学技术. 2024(01): 154-177 . 百度学术
4. 褚恒滨，高敏峰，王超，徐广华，王振平，张铎. 采煤工作面煤自燃标志气体CO的临界值研究. 煤矿现代化. 2024(02): 94-98 . 百度学术
5. 邓军，王津睿，任帅京，王彩萍，屈高阳，马砺. 采空区煤自燃高温点识别与探测技术研究与展望. 煤炭学报. 2024(02): 885-901 . 百度学术
6. 李夏，崔竟成. 煤矿井下外因火灾智能防控系统设计. 工矿自动化. 2024(S1): 100-103 . 本站查看
7. 周建. 基于“一通三防”的矿井智能管控技术研究与应用. 中国煤炭. 2024(06): 59-66 . 百度学术
8. 孙文亮，刘红亮，王飞，王海平，胡世熊. 煤矿密闭区域火情监测预警方法与管控体系研究. 煤. 2024(08): 30-35 . 百度学术
9. 李慎君，沈伟，李明建，覃木广，宋志强，杨宇. 全类型重大灾害监测预警系统架构——以宁煤麦垛山煤矿为例. 能源与环保. 2024(09): 57-64+70 . 百度学术
10. 高睿. 易自燃特厚煤层煤自燃预测预报系统研究与应用. 能源技术与管理. 2024(05): 143-145+198 . 百度学术
11. 王玉怀，胡硕鹏，朱永兴，窦静文，张烜乐，张思佳. 机器学习在煤自燃预测预报中的应用现状及展望. 中国煤炭. 2024(10): 98-103 . 百度学术
12. 杨峰峰，王海晖，龚恒敏，白光杰，张巨峰. 煤自燃倾向性测试及监测技术研究进展. 煤. 2023(02): 29-32+53 . 百度学术
13. 蒋恒. 煤自燃多参数气体主辅预测指标体系研究. 中国矿业. 2023(04): 147-153 . 百度学术
14. 孟清华，白亚娥，燕利芳，张奇勇. 浅埋近距离煤层群开采煤自燃发火防治技术研究. 内蒙古煤炭经济. 2023(03): 8-10 . 百度学术
15. 路培超，唐盛禹，蒋孝元，赵波. 瓦斯与煤自燃复合灾害监测预警系统研究与应用. 能源与环保. 2023(04): 10-15 . 百度学术
16. 降华. 基于云计算的煤炭自燃安全监测系统设计. 煤炭技术. 2023(08): 154-158 . 百度学术
17. 王俊博，李鑫，田继军，韦波，彭诗淇. 煤炭开发利用产业碳足迹计算方法及减排措施综述. 煤炭学报. 2023(S1): 263-274 . 百度学术
18. 葛俊岭，朱现峰，徐振. 回采面全周期火灾综合监测技术研究与应用. 现代矿业. 2023(08): 216-219 . 百度学术
19. 叶正亮，龚选平，尚博，胡冕. 煤自燃全过程精准预测预报指标体系研究与应用. 矿业研究与开发. 2023(09): 141-145 . 百度学术
20. 曹其嘉，陈辉，韦钊，杨博飞. 基于智能算法矿井多因素监测预警平台研究. 煤炭技术. 2023(10): 150-152 . 百度学术
21. 孔彪，朱思想，胡相明，杨涛，赵旭帅，斐达特，万姝含. 基于改进鲸鱼算法优化BP神经网络的煤自燃预测研究. 矿业安全与环保. 2023(05): 30-36 . 百度学术
22. 李伟宏. 雁南煤矿智能通风控制系统研究. 工矿自动化. 2023(S2): 66-70+74 . 本站查看
23. 王丹. RVM在煤自燃预测中的应用研究. 煤. 2022(04): 1-5 . 百度学术
24. 张立海，孙建卫，赵兵兵. 储煤仓预防自燃的实践. 选煤技术. 2022(02): 41-44 . 百度学术
25. 郭甘霖. 基于PLC的综采设备状态监测系统设计研究. 机械管理开发. 2022(06): 270-271 . 百度学术
26. 王民华，牛显. 基于数据重构增强的采空区遗煤自燃预测模型. 煤矿安全. 2022(09): 86-93 . 百度学术
27. 华坤鹏，逯园坤. 烟煤物理吸附氧气的吸附机理. 煤矿安全. 2022(11): 48-55 . 百度学术
其他类型引用(11)
计量
- 文章访问数: 37
- HTML全文浏览量: 7
- PDF下载量: 5
- 被引次数: 38
