-
Notifications
You must be signed in to change notification settings - Fork 0
/
Chapter5_PipeRecon.py
234 lines (199 loc) · 8.56 KB
/
Chapter5_PipeRecon.py
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
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
################################################
# This script produces figures for thesis chapter 5
# related to the case study of CT of subsea pipes
# By Silja L. Christensen
# May 2024
################################################
#%%
import numpy as np
import matplotlib.pyplot as plt
import funs
from cil.framework import AcquisitionGeometry, ImageGeometry, AcquisitionData, BlockDataContainer
from cil.plugins.astra.operators import ProjectionOperator
from cil.optimisation.operators import BlockOperator, IdentityOperator, GradientOperator
from cil.optimisation.algorithms import FISTA
from cil.optimisation.functions import LeastSquares, ZeroFunction
from cil.plugins.ccpi_regularisation.functions import FGP_TV
#%%
#=======================================================================
# Initialize
#=========================================================================
# Data: can be downloaded via https://zenodo.org/records/6817690
datapath = '../FORCE/data/Data_20180911/'
datafile = 'sinoN8.dat'
# Filepath for saving figures
path = 'figures/realrecon_geom{}_'.format(ag)
# Aqcuisition geometry type
# Choose "full", "limited90" or "sparseangles20percent"
ag = "limited90"
# Reconstruction domain discretization size
N = 500
# Noise std
data_std = 0.05
# Colormap for plotting
cmap = 'gray'
cmin = -0.05
cmax = 0.2
# Projection backend
device = 'gpu'
# Optimization parameters
if ag == "full":
noreg_iter = 100
alpha_1st_tikh = 0.8
tikh_1st_iter = 200
alpha_tikh = 3
tikh_iter = 100
alpha_TV = 0.1
TV_iter = 500
elif ag == "limited90":
noreg_iter = 100
alpha_1st_tikh = 0.8
tikh_1st_iter = 200
alpha_tikh = 3
tikh_iter = 100
alpha_TV = 0.1
TV_iter = 230
elif ag == "sparseangles20percent":
noreg_iter = 100
alpha_1st_tikh = 0.8
tikh_1st_iter = 200
alpha_tikh = 3
tikh_iter = 100
alpha_TV = 0.1
TV_iter = 230
#%%=======================================================================
# Define geometry
#=========================================================================
# Load aqusition geometry from library
det_count, angles, source_y, detector_y, beamshift_x, _, det_elem_size, det_length = funs.geom_Data20180911(ag)
angle_count = len(angles)
# Size of reconstruction domain
domain=(55, 55)
# Start view angle
if ag == "limited90":
start_angle = -np.pi/2-2*15/180*np.pi
elif ag == "sparseangles20percent":
start_angle = 4/180*np.pi
else:
start_angle = 0
# CIL geometries
cil_ag = AcquisitionGeometry.create_Cone2D(source_position = [beamshift_x,-source_y],
detector_position = [beamshift_x,detector_y],
detector_direction_x=[1, 0],
rotation_axis_position=[0, 0])
cil_ag.set_angles(angles=angles, initial_angle = start_angle, angle_unit='radian')
cil_ag.set_panel(num_pixels=det_count, pixel_size=det_elem_size)
cil_ig = ImageGeometry(voxel_num_x=N,
voxel_num_y=N,
voxel_size_x=domain[0]/N,
voxel_size_y=domain[0]/N)
#%%=======================================================================
# Load sinogram
#=========================================================================
# load data and change to correct data structure
sino = np.loadtxt(datapath + datafile, delimiter=';')
sino = sino.astype('float32')
sino_astra = np.rot90(sino, k = 1)
sino_astra = sino_astra[:-8, :]
if ag == 'sparseangles50percent':
sino_astra = sino_astra[::4, :]
elif ag == 'sparseangles20percent':
sino_astra = sino_astra[::10, :]
elif ag == 'sparseangles':
sino_astra = sino_astra[::20, :]
elif ag == 'full':
sino_astra = sino_astra[::2, :]
elif ag == 'limited90':
sino_astra = sino_astra[::2, :][15:105,:]
b_data = sino_astra.flatten()
b_data = b_data.reshape(sino_astra.shape).flatten(order="F")
data = AcquisitionData(array=np.rot90(b_data.astype(np.float32).reshape(det_count, angle_count)), geometry=cil_ag)
#%%=======================================================================
# Forward projector
#=========================================================================
# cil
A_cil = ProjectionOperator(cil_ig, cil_ag, device = device)
#%%=======================================================================
# CGLS reconstruction
#=========================================================================
F = LeastSquares(A_cil, data)
initial = cil_ig.allocate(0.0)
FISTA_noreg = FISTA(f = F, g = ZeroFunction(), initial=initial, max_iteration = 1000, update_objective_interval = 1)
FISTA_noreg.run(noreg_iter, verbose=True)
recon_noreg = FISTA_noreg.solution
# Figure 5.3, column 1, row 1/2/3 (ag="full"/"sparseangles20percent"/"limited90")
fig, ax = plt.subplots(nrows = 1, ncols = 1, figsize = (5,4))
cs = ax.imshow(np.rot90(recon_noreg.as_array(), k = 3), extent=[0, N, N, 0], aspect='equal', cmap=cmap, vmin = cmin, vmax = cmax)
cax = fig.add_axes([ax.get_position().x1+0.01,ax.get_position().y0,0.03,ax.get_position().height])
cbar = plt.colorbar(cs, cax=cax)
ax.tick_params(axis='both', which='both', length = 0, labelsize=12)
plt.setp(ax.get_xticklabels(), visible=False)
plt.setp(ax.get_yticklabels(), visible=False)
cbar.ax.tick_params(labelsize=12)
plt.savefig(path + 'noreg.png', transparent=False)
plt.close()
#%%=======================================================================
# 1st order Tikhonov reconstruction
#=========================================================================
F = LeastSquares(A_cil, data)
L = GradientOperator(cil_ig)
operator_block = BlockOperator(A_cil, alpha_1st_tikh*L)
zero_data = L.range.allocate(0)
data_block = BlockDataContainer(data, zero_data)
F_T = LeastSquares(operator_block, data_block)
#setup CGLS with the Block Operator and Block DataContainer
initial = cil_ig.allocate(0.0)
FISTA_1st_tikh = FISTA(f = F_T, g = ZeroFunction(), initial=initial, max_iteration = 1000, update_objective_interval = 1)
#run the algorithm
FISTA_1st_tikh.run(tikh_1st_iter, verbose = True)
recon_1st_Tik = FISTA_1st_tikh.solution
# Figure 5.3, column 2, row 1/2/3 (ag="full"/"sparseangles20percent"/"limited90")
fig, ax = plt.subplots(nrows = 1, ncols = 1, figsize = (5,4))
cs = ax.imshow(np.rot90(recon_1st_Tik.as_array(), k = 3), extent=[0, N, N, 0], aspect='equal', cmap=cmap, vmin = cmin, vmax = cmax)
cax = fig.add_axes([ax.get_position().x1+0.01,ax.get_position().y0,0.03,ax.get_position().height])
cbar = plt.colorbar(cs, cax=cax)
ax.tick_params(axis='both', which='both', length = 0, labelsize=12)
plt.setp(ax.get_xticklabels(), visible=False)
plt.setp(ax.get_yticklabels(), visible=False)
cbar.ax.tick_params(labelsize=12)
plt.savefig(path + '1st_Tik.png', transparent=False)
plt.close()
#%%=======================================================================
# TV reconstruction
#=========================================================================
F = LeastSquares(A_cil, data)
TV = FGP_TV(alpha=alpha_TV, device='gpu')
initial = cil_ig.allocate(0.0)
my_TV = FISTA(f=F,
g=TV,
initial=initial,
max_iteration=2000,
update_objective_interval=1)
my_TV.run(TV_iter, verbose=True)
recon_TV = my_TV.solution
# Figure 5.3, column 3, row 1/2/3 (ag="full"/"sparseangles20percent"/"limited90")
fig, ax = plt.subplots(nrows = 1, ncols = 1, figsize = (5,4))
cs = ax.imshow(np.rot90(recon_TV.as_array(), k = 3), extent=[0, N, N, 0], aspect='equal', cmap=cmap, vmin = cmin, vmax = cmax)
cax = fig.add_axes([ax.get_position().x1+0.01,ax.get_position().y0,0.03,ax.get_position().height])
cbar = plt.colorbar(cs, cax=cax)
ax.tick_params(axis='both', which='both', length = 0, labelsize=12)
plt.setp(ax.get_xticklabels(), visible=False)
plt.setp(ax.get_yticklabels(), visible=False)
cbar.ax.tick_params(labelsize=12)
plt.savefig(path + 'TV.png', transparent=False)
plt.close()
#%%=======================================================================
# Plot sinogram
#=========================================================================
# Figure 5.1(b)
fig, ax = plt.subplots(nrows = 1, ncols = 1, figsize = (6,4))
cs = ax.imshow(b_data.reshape(det_count,angle_count, order = 'C'), cmap=cmap, aspect=0.5)
cax = fig.add_axes([ax.get_position().x1+0.01,ax.get_position().y0,0.03,ax.get_position().height])
cbar = plt.colorbar(cs, cax=cax)
ax.tick_params(axis='both', which='both', labelsize=12)
cbar.ax.tick_params(labelsize=12)
ax.set_xlabel('Projection angle [deg]', fontsize = 12)
ax.set_ylabel('Detector pixel', fontsize = 12)
ax.set_xticks(np.linspace(0,angle_count,7, endpoint=True))
plt.savefig(path + 'sinogram.png', transparent=False)
plt.close()