-
Notifications
You must be signed in to change notification settings - Fork 1
/
Demo_Analysis_MS_CSNet.m
169 lines (148 loc) · 6.27 KB
/
Demo_Analysis_MS_CSNet.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
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% @article{Canh2018_MSCSNet,
% title={Multi-Scale Deep Compressive Sensing Network},
% author={Thuong, Nguyen Canh and Byeungwoo, Jeon},
% conference={IEEE International Conference on Visual Comunication and Image Processing},
% year={2018}
% }
% by Thuong Nguyen Canh (9/2018)
% ngcthuong@gmail.com
% https://github.com/AtenaKid
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% You need to install Matconvnet in order to run this code
warning('off','all')
addpath('D:\matconvnet-1.0-beta25\matlab\mex'); % Link to your Matconvnet Mex file
% addpath('matconvnet-1.0-beta25\matlab');
addpath('.\utilities');
folderTest = 'Classic13_512';
% networkTest = {'MS-CSNet1' 'MS-CSNet2' 'MS-CSNet3'}; % 10
networkTest = {'W-DCS1' 'W-DCS2' 'W-DCS3'}; % 102
showResult = 0;
writeRecon = 1;
featureSize = 64;
blkSize = 32;
isLearnMtx = [1, 0];
for netId = [1 2 3]
network = networkTest{netId};
for subRate = [0.1:0.1:0.3]
modelName = [network '_r' num2str(subRate)]; %%% model name
data = load(fullfile('models', network ,[modelName,'.mat']));
net = dagnn.DagNN.loadobj(data.net);
if strcmp(network,'CSNet')
net.renameVar('x0', 'input');
net.renameVar('x12', 'prediction');
else
net.removeLayer(net.layers(end).name) ;
end
net.mode = 'train';
net.move('gpu');
%%% read images
ext = {'*.jpg','*.png','*.bmp', '*.pgm', '*.tif'};
filePaths = [];
for i = 1 : length(ext)
filePaths = cat(1,filePaths, dir(fullfile('testsets',folderTest,ext{i})) );
end
PSNRs_CSNet = zeros(1,length(filePaths));
SSIMs_CSNet = zeros(1,length(filePaths));
count = 1;
allName = cell(1);
for i = 1:1:length(filePaths)
%%% read images
image = imread(fullfile('testsets', folderTest, filePaths(i).name));
[~,nameCur,extCur] = fileparts(filePaths(i).name);
allName{count} = nameCur;
if size(image,3) == 3
image = modcrop(image,32);
image = rgb2ycbcr(image);
image = image(:,:,1);
end
label = im2single(image);
if mod(size(label, 1), blkSize) ~= 0 || mod(size(label, 2), blkSize) ~= 0
continue
end
input = label;
input = gpuArray(input);
net.conserveMemory = false;
tic
net.eval({'input', input}) ;
time(i) = toc;
%% =========== Compressed Measurement ==========================
Y = squeeze(gather(net.vars(3).value));
edges = [-8:0.1:8];
his = histogram(Y(:), edges);
value = his.Values./sum(his.Values);
range = [min(Y(:)), max(Y(:))];
if ~exist(['histResults\' network '\']), mkdir(['histResults\' network '\']); end
save(['histResults\' network '\' nameCur '_r' num2str(subRate) '.mat'], 'his', 'value', 'range', 'Y');
% %% =========== Decomposition ===================================
% im0 = squeeze(gather(net.vars(2).value));
% im1 = im0(:, :, 1); im2 = im0(:, :, 2); im3 = im0(:, :, 3); im4 = im0(:, :, 4);
%
% im = cat(1, cat(2, im1, im2), cat(2, im3, im4)); imagesc(im);
% axis square;axis off; colormap(gray); colorbar; set(gca, 'fontsize', 20);
% im = export_fig; imwrite(im, ['Decomp\' modelName '_rate' num2str(subRate) '_grey.tif']);
%
%
% %% Check the sampling matrix
% Phi = squeeze(gather(net.params(1).value));
% tLegend = {'Low-Low', 'Low-High', 'High-Low', 'High-High'};
%
% [w, h, c, k] = size(Phi);
% k0 = floor(sqrt(k));
%
% bigMtx = [];
% for idx1 = 1:1:2
% tmpBig = [];
% for idx2 = 1:1:2
% idx = (idx1-1) * 2 + idx2;
%
% %pr = figure(idx);
% mtx = [];
% for i0 = 1:1:k0
% tmp = [];
% for i1 = 1:1:k0
% i2 = (i0 - 1)*k0 + i1;
% %pr = figure(1);
% Phi_i = Phi(:, :, idx, i2);
%
%
% tmp = [tmp, Phi_i];
% %min_i(i2) = min(Phi_i(:));
% %max_i(i2) = max(Phi_i(:));
% % disp(['range: [' num2str(min_i(i2)) ',' num2str(max_i(i2)) ']']);
%
% %im = [Phi_i(:, :, 1), Phi_i(:, :, 2); Phi_i(:, :, 3),Phi_i(:, :, 4)];
% %imshow(im, []);
% end
% mtx = [mtx; tmp];
% end
%
% tmpBig = [tmpBig, mtx] ;
%
% end
% bigMtx = [bigMtx; tmpBig];
% end
%
% pr = figure(1); imshow(bigMtx, []); set(gcf,'color','w'); set(gca, 'fontsize', 18);
%
% im = export_fig;
% imwrite(im, ['mtxResults_grey\meas_' modelName '_rate' num2str(subRate) '.tif']);
% saveas(pr, ['mtxResults_grey\meas_' modelName '_rate' num2str(subRate) '.png']);
% saveas(pr, ['mtxResults_grey\meas_' modelName '_rate' num2str(subRate) '.fig']);
% save(['mtxResults_grey\meas_' modelName '_rate' num2str(subRate) '.mat'], 'Phi');
%
% if subRate == 0.1
% imagesc(bigMtx, [-.2 .2]); axis square;axis off;
% colormap(gray); colorbar; set(gcf,'color','w');set(gca, 'fontsize', 26);
% elseif subRate == 0.2
% imagesc(bigMtx, [-0.25, 0.25]); axis square;axis off;
% colormap(gray); colorbar; set(gcf,'color','w');set(gca, 'fontsize', 26);
% end
% im = export_fig;
% imwrite(im, ['mtxResults_grey\meas_' modelName '_rate' num2str(subRate) '_grey.tif']);
% %% Check CS measurement
% %meas = net.getVarIndex('conv2') ;
% output = gather(squeeze(gather(net.vars(1).value))) ;
end
end
end