基于预滤波的ADMM图像恢复研究  

Pre-filtering-based ADMM Image Restoration

作　　者：彭施曼 牛俊瑶 张小会 马家庆 PENG Shiman;NIU Junyao;ZHANG Xiaohui;MA Jiaqing(School of Electrical Engineering,Guizhou University,Guiyang 550025,China;Department of Automation Engineering,Maotai Institute,Renhuai 564507,China)

机构地区：[1]贵州大学电气工程学院,贵州贵阳550025 [2]茅台学院酿酒工程自动化系,贵州仁怀564507

出　　处：《电工技术》2024年第6期88-90,共3页Electric Engineering

基　　金：贵州省科技厅支撑项目(黔科合支撑2023-G179)。

摘　　要：为了提高图像恢复的PSNR,在ADMM图像恢复算法的基础上,采用预滤波的方法进行研究。在三幅灰度图上通过施加均值滤波、中值滤波和小波重构三种预滤波处理,过程中分别改变计算的点数,结果显示均值滤波、中值滤波和小波重构的PSNR分别达到了28.95、33.24、38.58 dB,均比原ADMM算法有所提高,其中小波重构的效果最明显,表明该预滤波方法适用于ADMM的实际工程应用。In order to improve PSNR of image restoration,this paper adopts pre-filtering method on the basis of ADMM image restoration algorithm.Three kinds of pre-filtering are applied on three gray-scale images,namely mean filtering,median filtering and wavelet reconstruction,and the calculated points are changed independently in the process.The results show that PSNR of mean filtering,median filtering and wavelet reconstruction reaches 31.95 dB,33.24 dB and 38.58 dB respectively,which are improved compared with the original ADMM algorithm,and the effect of wavelet reconstruction is the most obvious.It shows that pre-filtering method in this paper is suitable for practical engineering application of ADMM.

关 键 词：预滤波 峰值信噪比 交替方向乘子算法 图像恢复 

分 类 号：TN911.73[电子电信—通信与信息系统]