欢迎访问《金属矿山》杂志官方网站,今天是 分享到:
×

扫码分享

金属矿山 ›› 2026, Vol. 55 ›› Issue (10): 214-220.

• • 上一篇    下一篇

矿井监控图像去噪的改进极小曲面滤波器

郭晶晶1 ,马云鹏2 ,王 锐1   

  1. 1.信阳学院大数据与人工智能学院,河南 信阳 464000;2.河海大学信息科学与工程学院,江苏 常州 213000
  • 出版日期:2026-11-03 发布日期:2026-11-09
  • 通讯作者: 马云鹏(1993—),男,副研究员,博士,硕士研究生导师。
  • 作者简介:郭晶晶(1989—),女,讲师,硕士。
  • 基金资助:
    国家自然科学基金青年科学基金项目(编号:62001156)。

Improved Minimal Surface Filter for Mine Monitoring Image Denoising

GUO Jingjing1 ,MA Yunpeng2 ,WANG Rui1   

  1. 1.School of Data and Artificial Intelligence,Xinyang University,Xinyang 464000,China; 2.College of Information Science and Engineering,Hohai University,Changzhou 213000,China
  • Online:2026-11-03 Published:2026-11-09

摘要: 井下复杂的光照条件易导致传统去噪算法难以区分真实场景细节与噪声,去噪后的图像容易出现过度 平滑或细节丢失的现象。为改善矿井监控图像的视觉质量,提出了基于改进极小曲面滤波器(Improved Minimal Sur face Filter,IMSF)的去噪方法。该方法首先通过引入动态噪声估计和自适应参数调整机制对极小曲面滤波器(Mini mal Surface Filter,MSF)进行改进,并结合Canny算子保护关键结构特征,同时在能量泛函中增加边缘保护约束项,以 避免边缘过度模糊。在此基础上,采用小波变换分析噪声分布特性,结合所提改进极小曲面滤波器实现矿井监控图 像去噪。试验结果表明:与小波阈值、非局部均值滤波(Non-local Means,NLM)及高斯滤波相比,通过IMSF对矿井监 控图像去噪后的峰值信噪比(Peak Signal-to-Noise Ratio,PSNR)与结构相似性指数(Structural Similarity Index,SSIM)最 高,当迭代200次时,PSNR与SSIM分别达到29.9 dB、0.97。在迭代次数不同的情况下,IMSF处理后的矿井监控图像 的信息保真度(Visual Information Fidelity,VIF)与局部对比度(Enhancement Measure Evaluation,EME)取值达到最优, 分别为28.4和0.81。研究反映出,采用IMSF实现矿井监控图像去噪效果较好,有助于改善矿井监控图像的视觉质量。

关键词: 矿井监控图像 , 图像去噪 , 改进极小曲面滤波器 , 小波变换 , Canny算子 , 自适应参数调整

Abstract: Complex underground lighting conditions make it difficult for traditional denoising algorithms to distinguish re al scene details from noise,and denoised images are prone to over-smoothing or loss of detail.To improve the visual quality of mine monitoring images,a denoising method based on an Improved Minimal Surface Filter (IMSF) is proposed.First,the Mini mal Surface Filter (MSF) is improved by introducing dynamic noise estimation and an adaptive parameter adjustment mecha nism,and the Canny operator is combined to protect key structural features.Meanwhile,an edge-preservation constraint term is added to the energy functional to avoid excessive edge blurring.On this basis,wavelet transform is used to analyze the noise distribution characteristics,and the proposed improved minimal surface filter is combined to denoise mine monitoring images. Test results show that,compared with wavelet thresholding,Non-local Means (NLM),and Gaussian filtering,the proposed IMSF achieves the highest Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) after denoising mine monitoring images.When the number of iterations is 200,the PSNR and SSIM reach 29.9 dB and 0.97,respectively.Under different iteration numbers,the Visual Information Fidelity (VIF) and Enhancement Measure Evaluation (EME) of mine mo nitoring images processed by IMSF reach the optimal values,which are 28.4 and 0.81,respectively.The study indicates that u sing IMSF to denoise mine monitoring images has a good effect and helps improve the visual quality of mine monitoring images.

Key words: mine monitoring image,image denoising,IMSF,wavelet transform,Canny operator,adaptive parameter adjust ment

中图分类号: