forked from Computational-Periscopy/Ordinary-Camera
-
Notifications
You must be signed in to change notification settings - Fork 0
/
fig_S1_S2.m
179 lines (133 loc) · 6.98 KB
/
fig_S1_S2.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
%% Plots multiple shifted reconstructions (Fig S1, S2)
% ------------------------------Pseudo-code------------------------------
% 1. Simulate light transport matrix A given scene geometry parameters.
%
% 2. Solve TV regularized optimization problem to reconstruct scene
% (FISTA) for 49 postulated occluder shifts, then combine using alpha
% trimmed mean.
% -----------------------------------------------------------------------
% Last Modified by Charles Saunders at Boston University
% 09-Nov-2018 (Clean-up and commented for sharing)
% Manuscript:
% Saunders, C. and Murray-Bruce, J and Goyal, V.K., 'Computational
% Periscopy with and Ordinary Digital Camera', Nature, 2018.
% Functions
addpath('Functions')
clear variables;
clc;
TestLetter = 'D11';
numPixels = 1008;
% Parameters
Ndiscr_mon = 6; %Discretization of each scene patch
downsamp_factor = 3; %Downsampling of measurements 2^downsamp_factor
viewAngleCorrection = 1; %True/False
useEstimatedOccPos = true; %Use estimated occluder position or not
load_experiment_config_data_reconstruction
% Data path
if ismac
calibParams.filepath = './Data/TestPosD11/';
elseif ispc
calibParams.filepath = '.\Data\TestPosD11\';
end
%%%%%%% Parameter setup %%%%%%%% (Nothing to manually set here)
% Wall/imaging plane
wall_point = [FOV_LLCorner(1) + FOV_size(1)/2,D,FOV_LLCorner(2)+FOV_size(2)/2]; %Point on plane
wall_vector_1 = [FOV_size(1)/2,0,0]; %Vector defining one direction of FOV (and extent)
wall_vector_2 = [0,0,FOV_size(2)/2]; %Vector defining the orthogonal direction (and extent)
wall_normal = cross(wall_vector_1,wall_vector_2);
wall_normal = wall_normal./norm(wall_normal);
walln_points = floor(numPixels/(2^downsamp_factor)); %Number of points to render in each direction
% Discretize imaging plane
f_imageplane = gpuArray((zeros(walln_points)));
wall_vec = (-1:2/(walln_points-1):1);
wall_matr(1,:) = gpuArray((wall_point(1) + wall_vec*wall_vector_1(1) + wall_vec*wall_vector_2(1)));
wall_matr(2,:) = gpuArray((wall_point(2) + wall_vec*wall_vector_1(2) + wall_vec*wall_vector_2(2)));
wall_matr(3,:) = gpuArray((wall_point(3) + wall_vec*wall_vector_1(3) + wall_vec*wall_vector_2(3)));
Monitor_xlim = [0 NumBlocks_col]*IlluminationBlock_Size(1) + Mon_Offset(1);
Monitor_y = 0;
Monitor_zlim = [0 NumBlocks_row]*IlluminationBlock_Size(2) + Mon_Offset(2);
Mon_xdiscr = linspace(Monitor_xlim(1),Monitor_xlim(2),NumBlocks_col);
Mon_zdiscr = linspace(Monitor_zlim(2),Monitor_zlim(1),NumBlocks_row);
wallparam.wall_matr = wall_matr;
wallparam.wall_point = wall_point;
wallparam.wall_vector_1 = wall_vector_1;
wallparam.wall_vector_2 = wall_vector_2;
wallparam.wall_normal = wall_normal;
wallparam.walln_points = walln_points;
numPixels = floor(numPixels/(2^downsamp_factor));
%% Load data
[test_image1,ground_truth1]=load_image1('image_test_mushroom20.mat',calibParams.filepath,downsamp_factor);
[test_image2,ground_truth2]=load_image1('image_test_smilehat20.mat',calibParams.filepath,downsamp_factor);
[test_image3,ground_truth3]=load_image1('image_test_colbar20.mat',calibParams.filepath,downsamp_factor);
[test_image4,ground_truth4]=load_image1('image_test_bur20.mat',calibParams.filepath,downsamp_factor);
%%
count = 1;
true_corner = Occ_LLcorner;
% Formula for calculating shifts in occluder position
scale = D/(D-Occ_LLcorner(2));
px_shiftx = IlluminationBlock_Size(1)/scale;
px_shifty = IlluminationBlock_Size(2)/scale;
%Step 4 pixels in steps of 2
xpos = [-6*px_shiftx, -4*px_shiftx,-2*px_shiftx, 0, 2*px_shiftx, 4*px_shiftx, 6*px_shiftx]; %Offsets of occluder to use
zpos = [-6*px_shifty, -4*px_shifty,-2*px_shifty, 0, 2*px_shifty, 4*px_shifty, 6*px_shifty];
pixel_shift_x = (xpos*scale)/IlluminationBlock_Size(1);
pixel_shift_y = (zpos*scale)/IlluminationBlock_Size(2);
pad = round([max(pixel_shift_x), max(pixel_shift_y)]);
tv_reg_param = 1.3*[0.05,0.05,0.05];
for zoff = 1:length(zpos)
for xoff = 1:length(xpos) % Loop through postulated occluder positions
%%%%%%%%%%%%%
Occ_LLcorner = true_corner + [xpos(xoff), 0, zpos(zoff)];
% Occluder
occ_corner(1,:,1) = Occ_LLcorner;
occ_corner(2,:,1) = Occ_LLcorner + [Occ_size(1), 0, 0];
occ_corner(3,:,1) = Occ_LLcorner + [Occ_size(1), 0, Occ_size(3)];
occ_corner(4,:,1) = Occ_LLcorner + [0, 0, Occ_size(3)];
occ_corner(1,:,2) = Occ_LLcorner + [Occ_size(1)/2-0.00275, 0, 0];
occ_corner(2,:,2) = Occ_LLcorner + [Occ_size(1)/2+0.00275, 0, 0];
occ_corner(3,:,2) = Occ_LLcorner + [Occ_size(1)/2-0.00275, 0, -Occ_LLcorner(3)];
occ_corner(4,:,2) = Occ_LLcorner + [Occ_size(1)/2+0.00275, 0, -Occ_LLcorner(3)];
%%%%%%%%%%%%%%
tic
[simA] = simulate_A(wallparam, occ_corner,simuParams, Mon_xdiscr,Mon_zdiscr, 0);
sr = 0.75*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sg = 0.83*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sb = 0.89*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
final_im1(:,:,:,count) = reconstruct_tv(sr*simA,sg*simA,sb*simA, test_image1, 3.25*tv_reg_param, NumBlocks_sim);
sr = 0.58*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sg = 0.72*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sb = 1.07*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
final_im2(:,:,:,count) = reconstruct_tv(sr*simA,sg*simA,sb*simA, test_image2, 3.25*tv_reg_param, NumBlocks_sim);
sr = 0.75*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sg = 0.75*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sb = 0.75*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
final_im3(:,:,:,count) = reconstruct_tv(sr*simA,sg*simA,sb*simA, test_image3, 3.5*tv_reg_param, NumBlocks_sim);
sr = 0.65*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sg = 0.85*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
sb = 0.95*125*125/(Ndiscr_mon^2)*prod(subblocksperaxis);
final_im4(:,:,:,count) = reconstruct_tv(sr*simA,sg*simA,sb*simA,test_image4, 3.25*tv_reg_param, NumBlocks_sim);
final_im1_t(:,:,:,count) = pad_translate(final_im1(:,:,:,count),[pixel_shift_x(xoff),pixel_shift_y(zoff)],pad);
final_im2_t(:,:,:,count) = pad_translate(final_im2(:,:,:,count),[pixel_shift_x(xoff),pixel_shift_y(zoff)],pad);
final_im3_t(:,:,:,count) = pad_translate(final_im3(:,:,:,count),[pixel_shift_x(xoff),pixel_shift_y(zoff)],pad);
final_im4_t(:,:,:,count) = pad_translate(final_im4(:,:,:,count),[pixel_shift_x(xoff),pixel_shift_y(zoff)],pad);
count = count + 1
toc
end
end
%%
for i=1:49
subplot(7,7,i)
imshow(cat(3,final_im1_t(:,:,1,i),final_im1_t(:,:,2,i),final_im1_t(:,:,3,i)))
end
for i=1:49
subplot(7,7,i)
imshow(cat(3,final_im2_t(:,:,1,i),final_im2_t(:,:,2,i),final_im2_t(:,:,3,i)))
end
for i=1:49
subplot(7,7,i)
imshow(cat(3,final_im3_t(:,:,1,i),final_im3_t(:,:,2,i),final_im3_t(:,:,3,i)))
end
for i=1:49
subplot(7,7,i)
imshow(cat(3,final_im4_t(:,:,1,i),final_im4_t(:,:,2,i),final_im4_t(:,:,3,i)))
end