def optimize_modifier(values, projections, geometry, samp = [1, 1, 1], key = 'axs_tan', metric = 'correlation', update = True, preview = False):
'''
Optimize a geometry modifier using a particular sampling of the projection array.
'''
maxiter = values.size
# Valuse of the objective function:
func_values = numpy.zeros(maxiter)
print('Starting a full search from: %0.3f' % values.min(), 'to %0.3f'% values.max())
time.sleep(0.3) # To print TQDM properly
ii = 0
for val in tqdm(values, unit = 'point'):
func_values[ii] = _modifier_l2cost_(projections, geometry, samp, val, key, metric = metric, preview = preview)
ii += 1
min_index = func_values.argmin()
display.plot2d(values, func_values, title = 'Objective: ' + key)
guess = _parabolic_min_(func_values, min_index, values)
if update:
geometry[key] = guess
print('Optimum found at %3.3f' % guess)
return guess
def equivalent_density(projections, geometry, energy, spectr, compound, density, preview = False):
'''
Transfrom intensity values to projected density for a single material array
'''
# Assuming that we have log array!
print('Generating the transfer function.')
if preview:
display.plot2d(energy, spectr, semilogy=False, title = 'Spectrum')
# Attenuation of 1 mm:
mu = model.linear_attenuation(energy, compound, density)
# Make thickness range that is sufficient for interpolation:
#m = (geometry['src2obj'] + geometry['det2obj']) / geometry['src2obj']
#img_pix = geometry['det_pixel'] / m
img_pix = geometry['img_pixel']
thickness_min = 0
thickness_max = max(projections.shape) * img_pix * 2
print('Assuming thickness range:', [thickness_min, thickness_max])
thickness = numpy.linspace(thickness_min, thickness_max, max(projections.shape))
exp_matrix = numpy.exp(-numpy.outer(thickness, mu))
synth_counts = exp_matrix.dot(spectr)
#flexUtil.plot(thickness, title = 'thickness')
#flexUtil.plot(mu, title = 'mu')
#flexUtil.plot(synth_counts, title = 'synth_counts')
if preview:
display.plot2d(thickness,synth_counts, semilogy=True, title = 'Attenuation v.s. thickness [mm].')
synth_counts = -numpy.log(synth_counts)
print('Callibration attenuation range:', [synth_counts[0], synth_counts[-1]])
print('array attenuation range:', [projections.min(), projections.max()])
print('Applying transfer function.')
time.sleep(0.5) # Give time to print messages before the progress is created
for ii in tqdm(range(projections.shape[1]), unit = 'img'):
projections[:, ii, :] = numpy.array(numpy.interp(projections[:, ii, :], synth_counts, thickness * density), dtype = 'float32')
return projections
def SIRT( projections, volume, geometry, iterations):
"""
Simultaneous Iterative Reconstruction Technique.
"""
ss = settings
preview = ss.preview
bounds = ss.bounds
logger.print('Feeling SIRTy...')
# Residual norms:
rnorms = []
# Progress bar:
pbar = _pbar_start_(iterations, 'iterations')
for ii in range(iterations):
# L2 update:
norm = l2_update(projections, volume, geometry)
if norm:
rnorms.append(norm)
# Apply bounds
if bounds:
numpy.clip(volume, a_min = bounds[0], a_max = bounds[1], out = volume)
if preview:
display.slice(volume, dim = 1, title = 'Preview')
else:
_pbar_update_(pbar)
# Stop progress bar
_pbar_close_(pbar)
if rnorms:
display.plot2d(rnorms, semilogy = True, title = 'Resudual L2')