当前位置:网站首页>【图像去噪】基于正则化实现图像去噪附matlab代码
【图像去噪】基于正则化实现图像去噪附matlab代码
2022-06-12 18:49:00 【Matlab科研工作室】
1 简介
图像去噪一直是图像处理问题中一个重要的备受关注的问题,降质图像对于图像的进一步应用造成了很大局限性.从数学角度上来讲,图像去噪是一个反问题,对于含有模糊核的图像处理而言更是一个不适定的反问题,目前解决反问题的最主要且行之有效的方法便是正则化方法.正则化方法的主要思想是引入逼近原不适定问题算子的正则化算子进而逼近原不适定问题的解,即通过正则化方法的求出的解是唯一的,存在的和稳定的且是原反问题的一个很好的逼近.
2 部分代码
%%%%梯度为0直接求解四种正则%%%%%%clear all;close all;x0 = double(imread('groundtruth\monarch.tif')); %导入groundtruthg0 = double(imread('degraded\monarch_n_0.1.tif')); %导入含噪图像g0k = fspecial('gaussian',19,2);%给定高斯模糊核lamda = 0.8; %正则化系数0.8[row column] = size(g0); %获取图像维度%图像二维傅里叶变换离散化K = fft2(k,row,column); %高斯核G = fft2(g0); %含噪图像%差分矩阵l1=[0 0 0;-1 1 0;0 0 0];l2=[0 -1 0;0 1 0;0 0 0];%傅里叶变换为频域530*530的矩阵Lx = fft2(l1,row,column);Ly = fft2(l2,row,column);%%%%%%%%%%使用二维傅里叶变换和梯度为零直接求解四种正则%%%%%%%%%%%L1正则:%x_iteration = ifft2((K'.*G-lamda)./(K'.*K));%L2正则:%x_iteration = ifft2(((K'.*G)./(K'.*K+2*lamda));%TV正则:%x_iteration = ifft2((K'.*G-lamda*Lx'-lamba*Ly')./(K'.*K));%TV-L2正则:x_iteration = ifft2((K'.*G)./(K'.*K-2*lamda*(Lx.^2)-2*lamda*(Ly.^2)));%%%%%%%%%%%显示重建图片%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%figure;colormap gray; %灰度化subplot(221);image(x0);title('GroundTruth');subplot(222);imagesc(g0);title('Degrade:0.1');subplot(224);imagesc(x_iteration);title('TV denoising:0.1');%%%%%%%%%%%%指标计算%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%[aa,bb] = size(x0);d_p = 9; %避免边界效应影响评估,去掉重建出来外面的一圈img1 = x_iteration(d_p:aa-d_p+1,d_p:bb-d_p+1); %重建图片img2 = x0(d_p:aa-d_p+1,d_p:bb-d_p+1); %groundtruthpsnr_result = psnr(img1,img2)[mssim, ~] = ssim_index(img1,img2)
3 仿真结果



4 参考文献
[1]李青青. 基于正则化方法的图像去噪模型的研究[D]. 武汉理工大学, 2013.
博主简介:擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真,相关matlab代码问题可私信交流。
部分理论引用网络文献,若有侵权联系博主删除。
边栏推荐
- Common troubleshooting tools and analysis artifacts are worth collecting
- tarfile解压嵌套tar
- Mise en œuvre de l'ACL réflexe dans le simulateur Cisco Cisco Packet Tracer
- 基于STM32设计智能家居控制系统(OneNet)_2022
- The practice of machine learning in meituan distribution system: restoring the real world with technology - Notes
- Leetcode topic [string] - Sword pointing offer 05- replace spaces
- leetcode:98. 统计得分小于 K 的子数组数目【双指针 + 计算子集个数 + 去重】
- Experiment 10 Bezier curve generation - experiment improvement - control point generation of B-spline curve
- 【历史上的今天】6 月 12 日:美国进入数字化电视时代;Mozilla 的最初开发者出生;3Com 和美国机器人公司合并
- leetcode:5270. 网格中的最小路径代价【简单层次dp】
猜你喜欢

Research Report on the overall scale, major manufacturers, major regions, products and application segmentation of swimming fins in the global market in 2022

Title 68: there are n integers, so that the previous numbers are moved backward m positions, and the last m numbers become the first m numbers

leetcode:6096. 咒语和药水的成功对数【排序 + 二分】

数据库全量SQL分析与审计系统性能优化之旅

Review of MySQL (IX): index

CVPR 2022 Oral 大连理工提出SCI:快速、超强的低光照图像增强方法

What is a network proxy

leetcode:5289. 公平分发饼干【看数据范围 + dfs剪枝】

Mise en œuvre de l'ACL réflexe dans le simulateur Cisco Cisco Packet Tracer

嵌入式开发:固件工程师的6项必备技能
随机推荐
Design of smart home control system (onenet) based on stm32_ two thousand and twenty-two
io.seata.common.exception.FrameworkException: can not connect to services-server.
Difference between rxjs of() and of ({})
CVPR 2022 Oral 大连理工提出SCI:快速、超强的低光照图像增强方法
MySQL - > > symbol usage JSON related
leetcode:98. Count the number of subarrays whose score is less than k [double pointers + number of calculated subsets + de duplication]
In 2021, the global spice and seasoning revenue is about 18720million US dollars, and it is expected to reach 25960million US dollars in 2028
Leetcode 416. Split equal sum subset
leetcode:6097. Match [set record + query one by one with the same length] after replacing characters
什么是网络代理
Regression analysis based on elasticnetcv
dumi 搭建文档型博客
On how to make digital transformation after the loan of large policy banks- Yixinhuachen
Six stone cognition: the apparent and potential speed of the brain
Getting started with the go language is simple: read / write lock
Go package import mode member visibility
Leetcode 416. 分割等和子集
用一个性能提升了666倍的小案例说明在TiDB中正确使用索引的重要性
Leetcode 494. 目标和
看不懂Kotlin源码?从Contracts 函数说起~