def calc_histograms_lab_setups():
for polimer_type in ["PDL-05", "PDLG-5002"]:
radius_coefs = {"PDL-05": 0.9, "PDLG-5002": 0.95}
paths = file_paths.get_benchtop_setup_paths(polimer_type)
for sample_id in range(len(paths)):
sample_name = list(paths.keys())[sample_id]
print(
f"============== {sample_id} sample: {sample_name} =============="
)
img3d = get_bin_img(sample_name)
print('tot: ', np.sum(img3d) / img3d.size)
fig, ax = plt.subplots()
ax.imshow(img3d[0], cmap="gray")
dm.save_plot(fig, "previews", f'{sample_id} bin ' + sample_name)
cylindric_fragments, cylindric_masks \
= divide_image_into_sector_cylindric_fragments(img3d,
height=len(img3d)//3-1,
radius_coef=radius_coefs[polimer_type])
fig = get_porosity_histogram_disrtibution(
cylindric_fragments,
sample_name,
img3d.shape,
pixel_size_mm=PIXEL_SIZE_MM_SETUP,
masks=cylindric_masks,
radius_coef=radius_coefs[polimer_type])
dm.save_plot(fig, SAVE_IMG_DESKTOP_SETUP_FOLDER,
f'hist {sample_id} {sample_name}')
def save_first_section_of_img(img_3d, file_id, edge_size):
with plt.style.context('classic'):
fig, ax = plt.subplots(figsize=(10, 10))
ax.imshow(img_3d[0], cmap='gray')
ax = plot_cubic_periodic_grid(img_3d[0], ax, edge_size)
ax.set_title(f'sample id {file_id}')
dm.save_plot(fig, SAVE_IMG_SYNCHROTON_FOLDER, f'section {file_id}')
def max_sections(data, folder, shots_num):
k = 0
img_frame = []
for img, _ in data:
img_frame.append(img)
k += 1
if k % shots_num == 0:
img = np.max(img_frame, axis=0)
fig, ax = plt.subplots(figsize=(7, 7))
ax.imshow(img, cmap="gray")
dm.save_plot(fig, folder, str(k // shots_num))
img_frame = []
def preview(sample_name, n, contrast=True, server=True):
if server:
img = dm.get_img2d_from_server(sample_name, n)
else:
file_name = dm.generate_tif_file_name(n, True)
img = dm.get_img2d_from_database(file_name,
folder_name="crop_"+sample_name)
if contrast:
clip_limit = 0.1
img = exposure.equalize_adapthist(img, clip_limit=clip_limit)
fig, ax = plt.subplots(figsize=(7, 7))
ax.imshow(img, cmap="gray")
dm.save_plot(fig, "sections", sample_name+"_"+str(n))
def plot_N_sections_multiotsu(img3d,
polimer_attenuation,
filename,
N=3,
radius_coef=0.9,
folder=SAVE_IMG_DESKTOP_SETUP_FOLDER):
section_indexes = np.linspace(0, len(img3d), N, endpoint=False, dtype=int)
fig, axes = plt.subplots(nrows=3, ncols=N, figsize=(21, 21))
axes_plot, axes_hist, axes_bin = axes
img3d = gaussian_filter(img3d, sigma=3)
for ax_plot, ax_hist, i in zip(axes_plot, axes_hist, section_indexes):
ax_plot.imshow(img3d[i], cmap="gray")
ax_hist.hist(img3d[i].flatten(), bins=255, color="gray")
thresholds = threshold_multiotsu(img3d[i])
for thresh in thresholds:
ax_hist.axvline(thresh, color='red')
img3d = binarize_without_eppendorf(img3d,
polimer_attenuation=polimer_attenuation)
img3d = lv.remove_levitating_stones(img3d)
for ax_bin, i in zip(axes_bin, section_indexes):
ax_bin.imshow(img3d[i], cmap="gray", interpolation=None)
center = np.asarray(img3d[i].shape) // 2
rr, cc = disk(center,
int(np.min(center) * radius_coef),
shape=img3d[i].shape)
mask = np.zeros(img3d[i].shape, dtype=int)
mask[rr, cc] = True
mask = np.ma.masked_where(mask > 0, mask)
ax_bin.imshow(mask, cmap="hsv", alpha=0.3)
dm.save_plot(fig, folder, 'section ' + filename)
img2d_mask = thresh(exposure.equalize_adapthist(img2d, clip_limit=clip_limit))
img2d_mask = get_small_pores_mask(img2d,
img2d_mask,
percentile_glob=97.5,
min_large_contour_length=1000,
window_size=200)
squares = [([500, 1500], [600,1500]),
([900, 1200], [900,1200])]
axes[0].imshow(img2d, cmap='gray')
viewer.view_applied_rectangle(img2d, *squares[0], axes[0], color='red')
viewer.view_applied_rectangle(img2d, *squares[1], axes[0], color='green')
axes[0].set_title("original image")
#axes[1].imshow(exposure.equalize_adapthist(img2d, clip_limit=0.05), cmap='gray')
viewer.view_applied_mask(exposure.equalize_adapthist(img2d, clip_limit=clip_limit), img2d_mask, axes[1], alpha=1)
viewer.view_applied_rectangle(img2d, *squares[0], axes[1], color='red')
viewer.view_applied_rectangle(img2d, *squares[1], axes[1], color='green')
axes[1].set_title("adaptive contrast")
viewer.view_region(exposure.equalize_adapthist(img2d, clip_limit=clip_limit), img2d_mask, axes[2], *squares[0])
axes[2].set_title("RED box zoomed", fontdict={'color': 'red'})
viewer.view_region(img2d, img2d_mask, axes[3], *squares[1], alpha=0.3)
axes[3].set_title("GREEN box zoomed", fontdict={'color': 'green'})
dm.save_plot(fig, "plots", "section")
# fig, axes = plt.subplots(ncols=2, figsize=(14, 7), constrained_layout=True)
center_coords)
# new approach
img_without_contours_frag = median(img_without_contours_frag)
min_brightness = np.min(img_without_contours_frag)
max_brightness = np.max(img_without_contours_frag)
local_thresh = min_brightness + (max_brightness -
min_brightness) * 0.5
if local_thresh > global_thresh:
local_thresh = global_thresh
bin_cropped_fragment = img_without_contours_frag > local_thresh
paste(mask_frame, bin_cropped_fragment, center_coords)
return ~mask_frame.astype(bool)
if __name__ == '__main__':
file_id = '123497'
num = np.random.randint(0, 2120) #100 # 320
img2d_gray = get_2d_slice_of_sample_from_database(num, file_id=file_id)
# fig = preview_small_pores_detection_by_fragment(img2d_gray, plots=8)
# pores_mask = get_small_pores_mask(img2d_gray)
# ic(np.sum(pores_mask)/pores_mask.size)
# fig = preview_small_pores_detection_by_fragment(img2d_gray, percentile=2)
fig = preview_small_pores_detection_full(img2d_gray, percentile=2)
dm.save_plot(fig, "previews", f"preview_small_pores{file_id}")
hists.append(hist)
hist_mean = np.mean(hists, axis=0)
hist_std = np.std(hists, axis=0) if size_of_sampling > 1 else 0
mean_of_distribution = count_mean_from_hist(hist_mean, bins)
ax.axvline(mean_of_distribution,
color="orange",
label=f"mean={mean_of_distribution:.2f} mkm")
median_of_distribution = count_median_from_hist(hist_mean, bins)
ax.axvline(median_of_distribution,
color="lawngreen",
label=f"median={median_of_distribution:.2f} mkm")
total_mean_vol = np.mean(total_mean_pore_volume_in_layers)
stats = dict(unit_name="mkm",
sample_size=size_of_sampling,
size_type=size_type,
min_x_value=cpe.recount_volumes_to_diameters(min_pore_volume * pixel_size_mkm) \
if size_type == "diameter" else min_pore_volume * pixel_size_mkm,
max_x_value=cpe.recount_volumes_to_diameters(max_pore_volume * pixel_size_mkm) \
if size_type == "diameter" else max_pore_volume * pixel_size_mkm,
total_mean_pore_volume_in_layers=cpe.recount_volumes_to_diameters(total_mean_vol) * \
pixel_size_mkm if size_type == "diameter" else total_mean_vol * pixel_size_mkm,
)
plot_error_hist(hist_mean, hist_std, bins, ax, stats)
ax.legend()
ax.grid()
dm.save_plot(fig, "plots", f"histogram {file_id} {size_type}")
paths = file_paths.get_benchtop_setup_paths(polimer_type)
df = dm.load_data("setup_culindric_porosities.csv")
#df = pd.DataFrame(columns = ['polimer_type', 'sample_number', 'date', 'mean', 'std'])
for sample_id in range(len(paths)):
sample_name = list(paths.keys())[sample_id]
print(
f"============== {sample_id} sample: {sample_name} ==============")
print(sample_name.split())
img3d = ~get_bin_img(sample_name)
print('tot: ', np.sum(img3d) / img3d.size)
fig, ax = plt.subplots()
ax.imshow(img3d[0], cmap="gray")
dm.save_plot(fig, "previews", f'{sample_id} bin ' + sample_name)
cylindric_fragments, cylindric_masks \
= divide_image_into_sector_cylindric_fragments(img3d,
height=len(img3d)//3-1,
radius_coef=radius_coefs[polimer_type])
std, mean = get_mean_porosity_and_std(cylindric_fragments,
cylindric_masks)
data_info = sample_name.split()
polimer, sample_number, date = data_info[0][:-2], data_info[0][
-1], data_info[-1]
df = df.append(
{
'polimer_type': polimer,
figsize=(15, 30),
constrained_layout=True)
axes = axes.flatten()
axes[0].imshow(img2d_gray, cmap="gray")
axes[1].imshow(img_matched, cmap="gray")
img_matched_bin = img_matched < threshold
_ = axes[2].hist(img_matched.flatten(), bins=255)
axes[2].axvline(x=threshold, color='red')
axes[3].imshow(img_matched_bin, cmap="gray")
mask = cpe.find_mask_longest_contours((img_matched > threshold),
filter_by_contour_length=True,
min_contour_length=max_contour_size,
max_number_of_contours=None)
image_with_contours = binary_fill_holes(mask)
axes[4].imshow(image_with_contours, cmap="gray")
imgl, _ = label(image_with_contours.astype(int))
axes[5].imshow(imgl, cmap="hsv")
#axes[6].imshow(mask, cmap="gray")
# axes[7].imshow(np.logical_and(image_with_contours, mask), cmap="gray")
for i, ax in enumerate(axes):
if i != 2 and i != 4:
ax.axis("off")
dm.save_plot(fig, 'hr_bin_plots', 'histogram transforms')
import file_paths
import data_manager as dm
if __name__ == '__main__':
polimer_type = ["PDL-05", "PDLG-5002"][1]
radius_coefs = {"PDL-05": 0.9, "PDLG-5002": 0.95}
paths = file_paths.get_benchtop_setup_paths(polimer_type)
for sample_id in range(len(paths)):
sample_name = list(paths.keys())[sample_id]
img3d_grey = h5py.File(paths[sample_name], 'r')['Reconstruction'][:]
img3d_bin = ~get_bin_img(sample_name)
N = 3
fig, axes = plt.subplots(nrows=2, ncols=N, figsize=(21, 21))
axes_grey, axes_bin = axes
section_indexes = np.linspace(50,
len(img3d_bin),
N,
endpoint=False,
dtype=int)
for ax_grey, ax_bin, i in zip(axes_grey, axes_bin, section_indexes):
ax_grey.imshow(img3d_grey[i], cmap="gray")
ax_bin.imshow(img3d_bin[i], cmap="gray")
dm.save_plot(fig, "setup bin section",
'section ' + str(sample_id) + ' ' + sample_name)
import data_manager as dm
import matplotlib.pyplot as plt
from skimage.transform import radon, iradon
angles = np.arange(0, 180, 1)
blobns = 2
porsty = 0.5
shape = 1000
# phantom = generator.blobs([shape, shape], porosity=1 - porsty, blobiness=blobns)
margin_per_cent = 0.1
phantom = np.ones((shape, shape), dtype=int)
fig, ax = plt.subplots(1, 1, figsize=(10, 10))
ax.imshow(phantom, cmap='gray')
dm.save_plot(fig, 'plots', 'phantom')
source_object = shape * np.sqrt(2) + 80
object_det = shape * np.sqrt(2) + 80
sinogram_parallel = radon(phantom, theta=angles, circle=False)
sinogram_fan = np.transpose(
ae.astra_fp_2d_fan(phantom,
angles,
source_object,
object_det,
detector_size=10000))
fig, ax = plt.subplots(1, 2)
ax[0].imshow(sinogram_fan, cmap='gray')
ax[0].set_title('sinogram_fan')
ax[1].imshow(sinogram_parallel, cmap='gray')
axes = np.transpose(axes)
for row_index, (axes_row, k) in enumerate(zip(axes, k_vals)):
img2d_restored, saved_energy = poganins_correction(img2d, k)
thresh = threshold_otsu(img2d_restored) * 0.7
if row_index == ncols // 2:
axes_row[0].imshow(img2d, cmap='gray')
else:
axes_row[0].axis("off")
axes_row[1].imshow(img2d_restored, cmap='gray')
axes_row[1].set_title(f"saved energy: {saved_energy:.5f}, k: {k}")
axes_row[2].imshow(img2d_restored > thresh, cmap='gray')
axes_row[2].set_title(f"saved energy: {saved_energy:.5f}, k: {k}")
dm.save_plot(fig, "unbiased_poganin", f"{file_id}")
# k_vals = np.arange(0.1, 10, 0.5)
# energies = []
# ic(k_vals)
# for k in k_vals:
# _, s = poganins_correction(img2d, k)
# energies.append(s)
#
# fig, ax = plt.subplots(figsize=(7, 7))
# ax.plot(k_vals, energies)
# ax.grid()
# dm.save_plot(fig, "unbiased_poganin", f"{file_id}_energies")
data_info = sample_name.split()
polimer, sample_number, date = data_info[0][:-2], data_info[0][
-1], data_info[-1]
img3d = get_bin_img(sample_name)
snow = ps.networks.snow(im=img3d, voxel_size=9)
proj = op.io.PoreSpy.import_data(snow)
net, geom = proj
pore_diameters, pore_distances = geom['pore.diameter'], geom[
'throat.length']
title = f"DIAMETERS: {sample_name}"
fig, mean_diameters, median_diameters, standard_dev_diameters = plot_hist(
pore_diameters, "DIAMETERS", sample_name)
dm.save_plot(fig, "networks", f"{sample_id} " + title)
title = f"DISTANCES: {sample_name}"
fig, mean_distances, median_distances, standard_dev_distances = plot_hist(
pore_distances, "DISTANCES", sample_name)
dm.save_plot(fig, "networks", f"{sample_id} " + title)
# df = df.append({'polimer_type': polimer_type,
# 'sample_number': sample_number,
# 'median_diameters': median_diameters,
# 'mean_diameters': mean_diameters,
# 'standard_dev_diameters': standard_dev_diameters,
# 'mean_distances': mean_distances,
# 'median_distances': median_distances,
# 'standard_dev_distances': standard_dev_distances}, ignore_index=True)
