def test_from_dct(self):
# read a microstructure from a DCT index.mat file
m = Microstructure.from_dct(data_dir=PYMICRO_EXAMPLES_DATA_DIR,
grain_file='t5_dct_cen_index.mat',
use_dct_path=False)
self.assertEqual(len(m.grains), 146)
self.assertEqual(m.voxel_size, 0.0014)
def test_from_dct(self):
# read a microstructure from a DCT index.mat file
m = Microstructure.from_dct(data_dir=PYMICRO_EXAMPLES_DATA_DIR,
grain_file='t5_dct_cen_index.mat',
use_dct_path=False)
m.autodelete = True
self.assertEqual(m.grains.nrows, 146)
def merge_dct_scans(scan_list,
samtz_list,
use_mask=False,
overlap=-1,
root_dir='.',
write_to_h5=True):
"""Merge two DCT scans.
This function build a `Microstructure` instance for each DCT scan and calls merge_microstructures.
The overlap can be deduced from the samtz values or specified directly.
:param list scan_list: a list with the two DCT scan names.
:param list samtz_list: a list with the two samtz value (the order should match the scan names).
:param bool use_mask: a flag to also merge the absorption masks.
:param int overlap: the value to use for the overlap if not computed automatically.
:param str root_dir: the root data folder.
:param bool write_to_h5: flag to write the result of the merging operation to an HDF5 file.
:return: A new `Microstructure` instance of the 2 merged scans.
"""
from pymicro.crystal.microstructure import Microstructure
import numpy as np
import os
import h5py
scan_shapes = [] # warning, shapes will be in (z, y, x) form
micros = []
for scan in scan_list:
scan_path = os.path.join(root_dir, scan, '5_reconstruction',
'phase_01_vol.mat')
with h5py.File(scan_path) as f:
scan_shapes.append(f['vol'].shape)
print(f['vol'].shape)
# figure out the maximum cross section
max_shape = np.array(scan_shapes).max(axis=0)[[2, 1, 0]]
for scan in scan_list:
# read microstructure for this scan
dct_analysis_dir = os.path.join(root_dir, scan)
print('processing scan %s' % scan)
micro = Microstructure.from_dct(data_dir=dct_analysis_dir)
print('voxel_size is {}'.format(micro.voxel_size))
# pad both grain map and mask
print('max shape is {}'.format(max_shape))
print('vol shape is {}'.format(micro.grain_map.shape))
offset = max_shape - micro.grain_map.shape
offset[2] = 0 # do not pad along Z
padding = [(o // 2, max_shape[0] - micro.grain_map.shape[0] - o // 2)
for o in offset]
print('padding is {}'.format(padding))
micro.grain_map = np.pad(micro.grain_map, padding, mode='constant')
print('has mask ? {}'.format(hasattr(micro, 'mask')))
if use_mask:
micro.mask = np.pad(micro.mask, padding, mode='constant')
elif hasattr(micro, 'mask'):
print('deleting mask attribute since we do not want to use it')
delattr(micro, 'mask')
micros.append(micro)
# find out the overlap region (based on the difference in samtz)
overlap_from_samtz = int(
(samtz_list[1] + scan_shapes[1][0] // 2 * micros[1].voxel_size) /
micros[1].voxel_size -
(samtz_list[0] - scan_shapes[0][0] // 2 * micros[0].voxel_size) /
micros[0].voxel_size)
print('vertical overlap deduced from samtz positions is %d voxels' %
overlap_from_samtz)
if overlap < 0:
overlap = overlap_from_samtz
print('using an actual overlap of %d voxels' % overlap)
# we have prepared the 2 microstructures, now merge them
merged_micro = Microstructure.merge_microstructures(micros,
overlap,
plot=True)
if write_to_h5:
# write the result
merged_micro.to_h5()
return merged_micro