-
Notifications
You must be signed in to change notification settings - Fork 80
/
Demo_real_application_denoising_and_deblurring_gray.m
202 lines (143 loc) · 6.76 KB
/
Demo_real_application_denoising_and_deblurring_gray.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
%==========================================================================
% This is the testing code of a special case of SRMD (scale factor = 1) for real image <denoising & deblurring>.
% There are two models, "SRMDx1_gray.mat" for grayscale image, "SRMDx1_color.mat"
% for color image. The models can do:
% 1. Deblurring. (The kernel is assumed to be Gaussian-like!!! For other kernels, you should re-train the model!)
% there are two types of kernels,
% including isotropic Gaussian (width range: [0.1, 3]),
% anisotropic Gaussian ([0.5, 8]).
% 2. Denoising. the noise level range is [0, 75].
% For denoising only, set "kerneltype = 1; kernelwidth = 0.1." (i.e., delta kernel)
%
%==========================================================================
% The basic idea of SRMD is to learn a CNN to infer the MAP of general SISR (with special case of sf=1), i.e.,
% solve x^ = arg min_x 1/(2 sigma^2) ||kx - y||^2 + lamda \Phi(x)
% via x^ = CNN(y,k,sigma;\Theta) or x^ = CNN(y,kernel,noiselevel;\Theta).
%
% There involves two important factors, i.e., blur kernel (k; kernel) and noise
% level (sigma; nlevel).
%
% For more information, please refer to the following paper.
% @article{zhang2017learningsrmd,
% title={Learning a Single Convolutional Super-Resolution Network for Multiple Degradations},
% author={Kai, Zhang and Wangmeng, Zuo and Lei, Zhang},
% year={2017},
% }
%
% If you have any question, please feel free to contact with <Kai Zhang ([email protected])>.
%
% This code is for research purpose only.
%
% by Kai Zhang (Nov, 2017)
%==========================================================================
% clear; clc;
format compact;
addpath('utilities');
imageSets = {'Hilbert','Set5','Set14','BSD68','BSD100','Urban100'}; % testing dataset
%%======= ======= ======= degradation parameter settings ======= ======= =======
% For real image 'Hilbert', some examples of degradation setting are given as follows.
% sf = 1; nlevel = 8; kerneltype = 1; kernelwidth = 0.2; % denoising only
% sf = 1; nlevel = 10; kerneltype = 1; kernelwidth = 0.5; % denoising and sharpening(deblurring)
%%======= ======= ======= ======= ======= ======= ======= ======= ======= =======
%% select testing dataset, use GPU or not, ...
setTest = imageSets([1]); %
showResult = 1; % 1, show results; 2, save restored images
pauseTime = 1;
useGPU = 1; % 1 or 0, true or false
method = 'SRMD';
folderTest = 'testsets';
folderResult = 'results';
if ~exist(folderResult,'file')
mkdir(folderResult);
end
%% scale factor (it is fixed to 1)
sf = 1; %{1}
%% load model with scale factor sf
folderModel = 'models';
load(fullfile(folderModel,['SRMDx',int2str(sf),'_gray.mat']));
%net.layers = net.layers(1:end-1);
net = vl_simplenn_tidy(net);
if useGPU
net = vl_simplenn_move(net, 'gpu') ;
end
%% degradation parameter (noise level and kernel) setting
%############################# noise level ################################
% noise level, from a range of [0, 75]
nlevel = 10; % [0, 75]
kerneltype = 1; % {1}
%############################### kernel ###################################
% there are tree types of kernels, including isotropic Gaussian,
% anisotropic Gaussian, and estimated kernel k_b for isotropic Gaussian k_d
% under direct downsampler (x2 and x3 only).
if kerneltype == 1
% type 1, isotropic Gaussian---although it is a special case of anisotropic Gaussian.
kernelwidth = 0.5; % from a range of [0.1, 3]. set kernelwidth from (0.001, 0.2) to generate delta kernel (no blur)
kernel = fspecial('gaussian',15, kernelwidth); % Note: the kernel size is fixed to 15X15.
tag = ['_',method,'_x',num2str(sf),'_itrG_',int2str(kernelwidth*10),'_nlevel_',int2str(nlevel)];
elseif kerneltype == 2
% type 2, anisotropic Gaussian
nk = 1;%randi(size(net.meta.AtrpGaussianKernel,4)); % randomly select one
kernel = net.meta.AtrpGaussianKernel(:,:,:,nk);
tag = ['_',method,'_x',num2str(sf),'_atrG_',int2str(nk),'_nlevel_',int2str(nlevel)];
end
%##########################################################################
surf(kernel) % show kernel
view(45,55);
title('Assumed kernel');
xlim([1 15]);
ylim([1 15]);
pause(2)
close;
%% for degradation maps
global degpar;
degpar = single([net.meta.P*kernel(:); nlevel(:)/255]);
for n_set = 1 : numel(setTest)
%% search images
setTestCur = cell2mat(setTest(n_set));
disp('--------------------------------------------');
disp([' ----',setTestCur,'-----Super-Resolution-----']);
disp('--------------------------------------------');
folderTestCur = fullfile(folderTest,setTestCur);
ext = {'*.jpg','*.png','*.bmp'};
filepaths = [];
for i = 1 : length(ext)
filepaths = cat(1,filepaths,dir(fullfile(folderTestCur, ext{i})));
end
%% prepare results
folderResultCur = fullfile(folderResult, [setTestCur,tag]);
if ~exist(folderResultCur,'file')
mkdir(folderResultCur);
end
%% perform denoising or/and deblurring (only support Gaussian-like kernel)
for i = 1 : length(filepaths)
input = imread(fullfile(folderTestCur,filepaths(i).name));
%label = modcrop(label, 2);
[h,w,C] = size(input);
input = im_pad(input);
if C == 3
input = rgb2gray(input); % input = rgb2ycbcr(input); input = input(:,:,1); % another option
end
[~,imageName,ext] = fileparts(filepaths(i).name);
input1 = im2single(input);
%tic
if useGPU
input1 = gpuArray(input1);
end
res = vl_srmd(net, input1,[],[],'conserveMemory',true,'mode','test','cudnn',true);
%res = vl_srmd_concise(net, input1); % a concise version of "vl_srmd".
%res = vl_srmd_matlab(net, input1); % You should also set "useGPU = 0;" and comment "net = vl_simplenn_tidy(net);"
output = im2uint8(gather(res(end).x));
%toc;
output = im_crop(output,h,w);
input = im_crop(input,h,w);
%output2 = 0.8*output + 0.2*input; % add noise and structure back to make the output more visually plausible. or GAN?
disp([setTestCur,' ',int2str(i),' ',filepaths(i).name]);
if showResult
imshow(cat(2,input,output));
drawnow;
title(['Denoising and deblurring ',filepaths(i).name],'FontSize',12)
pause(pauseTime)
imwrite(output,fullfile(folderResultCur,[imageName,'_x',int2str(sf),'.png']));% save results
end
end
end