-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy pathmain_abswp_simulation.m
147 lines (129 loc) · 5.41 KB
/
main_abswp_simulation.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
% ------------------------------------------------------------------------
% SPIRIT main script. It simulates the acquisition and reconstruction of
% an image by a single-pixel camera. It implements the adaptive basis
% scan described in [1]. Pattern positivity is addressed using either
% pattern splitting or solving the pattern generalization problem as
% described in [2].
%
% The spirit parameters are defined in the 'opt' structure. Type 'help
% spiritopt' for a description of the fields of this structure.
% References
% [1] F. Rousset et al., "Adaptive basis scan by wavelet prediction for
% single-pixel imaging", IEEE Transactions on Computational Imaging, 3,
% 36-46, 2017.
% http://dx.doi.org/10.1109/TCI.2016.2637079
%
% [2] F. Rousset et al., "A semi nonnegative matrix factorization
% technique for pattern generalization in single-pixel imaging", IEEE
% Transactions on Computational Imaging,
% https://doi.org/10.1109/TCI.2018.2811910
% Author: F. Rousset, N. Ducros
% Institution: Creatis laboratory, University of Lyon, France
% Date: 30 Apr 2018
% Toolbox: SPIRiT 2.0 https://github.com/nducros/SPIRIT
% License: CC-BY-SA 4.0 https://creativecommons.org/licenses/by-sa/4.0/
% ------------------------------------------------------------------------
set(0,'DefaultFigureColor',[1 1 1]); % White background
set(0,'DefaultFigureUnits','Normalized'); % Normalized units
set(0,'DefaultFigurePosition',[0.1 0.1 0.7 0.7]); % Almost full screen
clear;
%close all;
clc;
%% Set Matlab Search path (just once)
wavelabdir = 'D:\Creatis\Programmes\Matlab\Wavelab850'; % Set YOUR path here!
path(fullfile(genpath(wavelabdir)),path);
path([pwd filesep 'function'],path);
%% I - Image or dataset
%- data_set: either an image (simulation) or an experimental mat-file.
%- If the latter is chosen, only the set of percentages p can be changed.
opt.datadir = 'data';
% opt.dataset = 'Jaszczak_CCD_128x128.png';
opt.dataset = 'Jaszczak_CCD_64x64.png';
%% II - Simulating single-pixel camera
%- The following parameters ARE used if an image is provided to abswp.
%- The following parameters are NOT used if wavelet coefficients
%- (experimental or simulated data) are provided to abswp.
%- Patterns directory
opt.patdir = 'pattern';
%- DMD Patterns dynamic.
opt.b = 8;
%- Photon flux emitted by the object (in ph/s)
opt.n0 = 2100; %
%- Photon flux DC component (in ph/s)
opt.alpha = 20000; % (in ph/s)
%- Collection time (in s)
opt.dt = 1;
%- Apply Poisson noise to measurement (true) or not (false)
opt.noise = true;
% opt.noise = false;
%% III - ABS-WP parameters
%- wavelet_name: orthogonal or biorthogonal wavelets can be used.
%- Choices are: 'Haar' for Haar wavelet
%- 'Db' for Daubechies
%- 'Battle' for Battle-Lemarie
%- 'LeGall' for CDF 2.2 (5/3) biorthogonal wavelet
%- 'CDF' for CDF 4.4 (9/7) biorthogonal wavelet
%- For other wavelets, see Get_Wavelet_Parameters.m.
% opt.wav = 'Haar';
opt.wav = 'Db';
%- par: additional parameter for Daubechies or Battle wavelets giving the
%- number of vanishing moments.
%- Choices are: 2,3,4,5,6,7,8,9,10 for Daubechies
% 2,4,6 for Battle
opt.par = 4;
%- p = [p_J, ... , p_1] specifies the percentage of the detail
%- coefficients that are kept at each scale j = 1 ... J. Note that the
%- decomposition level J is chosen through the length of p, i.e., J = length(p).
%- Basis scan examples (all p_j set to 1 or 0)
%- p = [1 1 0]; % J = 3, CR = 75%
%- p = [1 1 1 0]; % J = 4, CR = 75%
%- p = [1 1 1 1 0]; % J = 5, CR = 75%
%- Adaptive basis scan examples (At least one p_j s.t. 0 < p_j < 1)
%- p = [0.7 0.15 0.005 0]; % J = 4, CR = 98%
%- p = [0.85 0.5 0.05 0.0045]; % J = 4, CR = 95%
%- p = [0.9 0.75 0.21 0.015]; % J = 4, CR = 90%
%- p = [0.9 0.8 0.45 0.019]; % J = 4, CR = 85%
%- p = [0.9 0.8 0.71 0.02]; % J = 4, CR = 80%
%- p = [0.95 0.85 0.84 0.05]; % J = 4, CR = 75%
%- p = [0.95 0.85 0.77 0.4]; % J = 4, CR = 50%
% opt.p = [0.9 0.75 0.21 0.015]; % J = 4, CR = 90%
% opt.p = [0.9 0.8 0.45 0.019]; % J = 4, CR = 85%
opt.p = [0.95 0.85 0.84 0.05]; % J = 4, CR = 75%
% opt.p = [0.95 0.85 0.77 0.4]; % J = 4, CR = 50%
%% IV - Handling experimental contraints (choose pattern spliting OR SNMF)
%- Pattern splitting
opt.exp = 'split';
%- SNMF-based pattern generalization
% opt.exp = 'snmf';
% opt.batch = 10;
% opt.maxitr = 20000;
% opt.epsilon = 1e-5;
%% V - ABS-WP. Adaptive basis scan by wavelet prediction
[F,F_WT,opt] = readcheckdataset(opt);
[F_rec,F_WT_rec,M] = abswp(F,[],opt); % Note the second argument is empty
%% Image quality metric
PSNR_rec = psnr(F, F_rec);
%% Plot
CR = 100*cr(opt.p, size(F,1));
%
figure;
h1 = subplot(221);
imagesc(M), axis image,
title(sprintf('Locations of the acquired coefficients\nCR = %0.2f %%',CR));
colormap(h1,hot);
%
h2 = subplot(222);
imagesc(F_WT_rec), colorbar, axis image,
title(sprintf('Acquired %s coefficients\nDecomposition level J = %d',...
opt.wav, length(opt.p)));
colormap(h2,jet);
%
h3 = subplot(223);
imagesc(F), colorbar, axis image,
title('Ground truth (CCD) image');
colormap(h3,gray);
%
h4 = subplot(224);
imagesc(F_rec), colorbar, axis image,
title(sprintf('Image recovery using ABS-WP with %s\nPSNR_{rec} = %0.2f dB',opt.wav,PSNR_rec));
colormap(h4,gray);