DENOISEDEMO
denoising demo using shearlets
Contents
Description
Denoising demo using shearlet transform
Copyright
Copyright (C) 2011. Xiaosheng Zhuang, University of Osnabrueck
clc clear all; close all;
Reading image
1. image bridge img = imread('bridge.jpg'); img = img(89:600,1:512,:); 2. image barbara
img = imread('barbara.gif'); img1 = double(img(:,:,1)); %img = rgb2gray(img); %img = im2double(img); %img1= img1(1:2:512,1:2:512); %imshow(uint8(img1)); % size(img1)
generating a noisy image
sigma = 30; % noise level noise = randn(size(img1))*sigma; % Gaussian noise noisyImg = img1+noise;
Setting parameters
R = 2; % oversammpling rate R = 2, 4, 8, 16; beta = 2; % scaling factor beta = 2, 4 Choice = 1; % basis Choice, see GENERATEW CG = 0; % Use CG or not err = 1e-5; % CG control error its = 10; % CG maximal iterations thr = 0.01*sigma; % setting threshold
Denoising
tic recImg = dshden(noisyImg,thr,R,beta,Choice,0,CG,err,its); toc
Forward Shearlet Transform... ppFT ... Weighting ... Windowing ... Decomposition of high-frequency part... scale = 9, Number of Tiles = 33 scale = 8, Number of Tiles = 33 scale = 7, Number of Tiles = 17 scale = 6, Number of Tiles = 17 scale = 5, Number of Tiles = 9 scale = 4, Number of Tiles = 9 scale = 3, Number of Tiles = 5 scale = 2, Number of Tiles = 5 scale = 1, Number of Tiles = 3 scale = 0, Number of Tiles = 3 Decomposition of low frequency part... Done! Done with Forward Shearlet Transform! Backward Shearlet Transform... Adjoint Windowing... Reconstruction from high-frequency part... scale = 9, Number of Tiles = 33 scale = 8, Number of Tiles = 33 scale = 7, Number of Tiles = 17 scale = 6, Number of Tiles = 17 scale = 5, Number of Tiles = 9 scale = 4, Number of Tiles = 9 scale = 3, Number of Tiles = 5 scale = 2, Number of Tiles = 5 scale = 1, Number of Tiles = 3 scale = 0, Number of Tiles = 3 Reconstrucion from low frequency part... Done! Weighting... AdjppFT... Done with Backward Shearlet Transform! Elapsed time is 4.030394 seconds.
dispay original image, noisy image, and denoised image
psnr1 = PSNR(img1,noisyImg); figure(1), imshow(uint8(img1)); figure(2), imshow(uint8(noisyImg)); text(10,-15,['PSNR = ' num2str(psnr1)]); psnr2 = PSNR(img1,real(recImg)) ; figure(3),imshow(uint8(real(recImg)));text(10,-15,['PSNR = ' num2str(psnr2)]);
