基于自组织迁移算法小波阈值矿区遥感图像去噪研究  

Study on Denoising of Mining Area Remote Sensing Images Based on Self-organizing Transfer Algorithm and A Wavelet Threshold

作　　者：胡志峰 吴龙华 陈竹安[2] HU Zhifeng;WU Longhua;CHEN Zhuan(Jiangxi Industrial Engineering Vocational and Technical College,337000,Pingxiang,Jiangxi,PRC;School of Surveying and Geoinformation Engineering,East China University of Technology,330013,Nanchang,PRC)

机构地区：[1]江西工业工程职业技术学院,江西萍乡337000 [2]东华理工大学测绘与空间信息工程学院,南昌330013

出　　处：《江西科学》2025年第1期59-64,共6页Jiangxi Science

基　　金：江西省教育厅科学技术研究项目(GJJ2206614);教育部人文社会科学研究项目(22YJCZH150);江西省社科基金“十四五”地区项目(22DQ44)。

摘　　要：遥感图像在矿区环境保护、矿山生态修复、地质情况监测等起到了重要作用,矿区环境情况复杂,遥感图像在获取的过程中噪声存在会严重影响图像地物分类的准确性。为此,在已有阈值函数的基础上构建基于自组织迁移算法的小波阈值函数,通过自组织迁移算法(SOMA)对函数参数进行多路径最优值的筛选。再利用峰值信噪比(PSNR)、均方误差(MSE)、结构相似度(SSIM)指数进行图像去噪评价,结果表明,通过数学分析改进阈值函数具有较好连续性,基于自组织算法的阈值函数对于矿山遥感图像去噪质量优于传统的硬阈值,软阈值函数。所改进的新阈值函数对于矿山图像的地物判别也有进一步提高。Remote sensing images play an important role in environmental protection,ecological restoration,and geological monitoring in mining areas.The environmental conditions in mining areas are complex,and the presence of noise in the acquisition process of remote sensing images can seriously affect the accuracy of image land classification.To address this issue,a wavelet threshold function based on self-organizing transfer alg orithm was developed,building on the existing threshold functions.Self-organizing transfer alg orithm(SOMA)was used to screen the function parameters for multi-path optimal values.The denoisin g performance was evaluated usin gpeak signal-to-noise ratio(PSNR),mean square error(MSE),and structural similarity(SSIM)index.The results show that improving the threshold function through mathematical analysis has good continuity,and the threshold function based on self-organizing algorithm has better denoising quality for mine remote sensing images than traditional hard threshold and soft threshold functions.Additionally,the improved threshold function also further enhances the discrimination of land features in mining images.

关 键 词：矿山遥感图像 SOMA算法 小波变换 阈值函数模型 

分 类 号：P237[天文地球—摄影测量与遥感]