基于小波变换和均值滤波的图像去噪方法  

Image Denoising Method Based on Wavelet Transform and Mean Filtering

作　　者：梁利利 高楠 李建军[2] LIANG Lili;GAO Nan;LI Jianjun(College of Foreign Languages,Xianyang Normal Unviersity,Xianyang 712000;Northern Beijing Vocational Education Institute,Beijing 101400)

机构地区：[1]咸阳师范学院外国语学院,咸阳712000 [2]北京京北职业技术学院,北京101400

出　　处：《计算机与数字工程》2019年第5期1229-1232,共4页Computer & Digital Engineering

基　　金：咸阳师范学院2015年科研立项课题"高等院校开放型实验室统筹管理模式研究"(编号:15XSYK005);咸阳师范学院2017年科研立项课题"从顺应论看习近平主席外访演讲中的语用移情"(编号:XSYK17024)资助

摘　　要：通过对均值滤波缺点和小波变换优势的分析,特提出基于小波变换和均值滤波的图像去噪方法。首先对含噪图像进行基于小波变换的阈值去噪处理,将处理后的小波系数通过小波变换实现局部重构,提取低频近似图像、水平、垂直、和对角三个部分的高频细节,并针对含噪图像的特点,使低频部分保持不变,对三个高频细节部分选用适合的均值滤波模板进行去噪处理,最后将近似低频细节和去噪后的三个高频信号进行小波重构,得到最终去噪后的图像。实验结果表明,该方法能在降低噪声的同时较好地保留图像细节,比单一使用这两种去噪方法具有更好的去噪效果。This paper proposes the image denoising method on the basis of the disadvantages of the wavelet transform and the advantages of mean filtering. Firstly,based on wavelet transform,threshold denoising of the noisy image is processed. And then the following wavelet coefficients are used to realize local reconstruction through wavelet transform,and extract the high frequency de. tails of low-frequency approximate images,horizontal,vertical,and diagonal parts. In the light of the characteristics of the noisy im. age,the low frequency part remains the same,and the suitable mean filtering template for the three high frequency details is select. ed. Finally,the low-frequency details and the three high-frequency signals are reconstructed and the image is finally denoised. The experimental result shows that the method can reduce noise and retain the image detail better. Using two methods of denoising meth. ods has better denoising effect than using one of them.

关 键 词：图像去噪 小波变换 均值滤波 滤波模板 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]