基于预滤波的ADMM图像恢复研究
Pre-filtering-based ADMM Image Restoration
彭施曼 1牛俊瑶 2张小会 2马家庆2
作者信息
- 1. 贵州大学电气工程学院,贵州 贵阳 550025;茅台学院酿酒工程自动化系,贵州 仁怀 564507
- 2. 贵州大学电气工程学院,贵州 贵阳 550025
- 折叠
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.
Abstract
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 re-sults 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 reconstruc-tion 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Key words
pre-filtering/PSNR/ADMM/image restoration引用本文复制引用
基金项目
贵州省科技厅支撑项目(黔科合支撑2023-G179)
出版年
2024
NETL
NSTL
万方数据