def __display_image(self,filename):
lip = LoadImagesImageProvider("dummy", "", filename, True)
image = lip.provide_image(None).pixel_data
frame = FIG.CPFigureFrame(self.__list_box.GetTopLevelParent(),
title = filename,
subplots=(1,1))
if image.ndim == 3:
frame.subplot_imshow(0,0,image,filename)
else:
frame.subplot_imshow_grayscale(0,0,image,filename)
frame.Refresh()
def __display_image(self, filename):
lip = LoadImagesImageProvider("dummy", "", filename, True)
image = lip.provide_image(None).pixel_data
frame = FIG.CPFigureFrame(self.__list_box.GetTopLevelParent(),
title=filename,
subplots=(1, 1))
if image.ndim == 3:
frame.subplot_imshow(0, 0, image, filename)
else:
frame.subplot_imshow_grayscale(0, 0, image, filename)
frame.Refresh()
def estimate_absorbance(self):
"""Load an image and use it to estimate the absorbance of a stain
Returns a 3-tuple of the R/G/B absorbances
"""
from cellprofiler.modules.loadimages import LoadImagesImageProvider
import wx
dlg = wx.FileDialog(None, "Choose reference image",
cpprefs.get_default_image_directory())
dlg.Wildcard = (
"Image file (*.tif, *.tiff, *.bmp, *.png, *.gif, *.jpg)|"
"*.tif;*.tiff;*.bmp;*.png;*.gif;*.jpg")
if dlg.ShowModal() == wx.ID_OK:
lip = LoadImagesImageProvider("dummy", "", dlg.Path)
image = lip.provide_image(None).pixel_data
if image.ndim < 3:
wx.MessageBox(
"You must calibrate the absorbance using a color image",
"Error: not color image",
style=wx.OK | wx.ICON_ERROR,
)
return None
#
# Log-transform the image
#
eps = 1.0 / 256.0 / 2.0
log_image = np.log(image + eps)
data = [-log_image[:, :, i].flatten() for i in range(3)]
#
# Order channels by strength
#
sums = [np.sum(x) for x in data]
order = np.lexsort([sums])
#
# Calculate relative absorbance against the strongest.
# Fit Ax = y to find A where x is the strongest and y
# is each in turn.
#
strongest = data[order[-1]][:, np.newaxis]
absorbances = [lstsq(strongest, d)[0][0] for d in data]
#
# Normalize
#
absorbances = np.array(absorbances)
return absorbances / np.sqrt(np.sum(absorbances**2))
return None
def estimate_absorbance(self):
'''Load an image and use it to estimate the absorbance of a stain
Returns a 3-tuple of the R/G/B absorbances
'''
from cellprofiler.modules.loadimages import LoadImagesImageProvider
import wx
dlg = wx.FileDialog(
None, "Choose reference image",
cpprefs.get_default_image_directory())
dlg.Wildcard = ("Image file (*.tif, *.tiff, *.bmp, *.png, *.gif, *.jpg)|"
"*.tif;*.tiff;*.bmp;*.png;*.gif;*.jpg")
if dlg.ShowModal() == wx.ID_OK:
lip = LoadImagesImageProvider("dummy", "", dlg.Path)
image = lip.provide_image(None).pixel_data
if image.ndim < 3:
wx.MessageBox("You must calibrate the absorbance using a color image",
"Error: not color image",
style=wx.OK | wx.ICON_ERROR)
return None
#
# Log-transform the image
#
eps = 1.0 / 256.0 / 2.0
log_image = np.log(image + eps)
data = [- log_image[:, :, i].flatten() for i in range(3)]
#
# Order channels by strength
#
sums = [np.sum(x) for x in data]
order = np.lexsort([sums])
#
# Calculate relative absorbance against the strongest.
# Fit Ax = y to find A where x is the strongest and y
# is each in turn.
#
strongest = data[order[-1]][:, np.newaxis]
absorbances = [lstsq(strongest, d)[0][0] for d in data]
#
# Normalize
#
absorbances = np.array(absorbances)
return absorbances / np.sqrt(np.sum(absorbances ** 2))
return None
def run(self, workspace):
statistics = []
m = workspace.measurements
assert isinstance(m, cpmeas.Measurements)
#
# Hack: if LoadSingleImage is first, no paths are populated
#
if self.file_wants_images(self.file_settings[0]):
m_path = "_".join(
(C_PATH_NAME, self.file_settings[0].image_name.value))
else:
m_path = "_".join((C_OBJECTS_PATH_NAME,
self.file_settings[0].objects_name.value))
if m.get_current_image_measurement(m_path) is None:
self.prepare_run(workspace)
image_set = workspace.image_set
for file_setting in self.file_settings:
wants_images = self.file_wants_images(file_setting)
image_name = (file_setting.image_name.value
if wants_images else file_setting.objects_name.value)
m_path, m_file, m_md5_digest, m_scaling, m_height, m_width = [
"_".join((c, image_name)) for c in (
C_PATH_NAME if wants_images else C_OBJECTS_PATH_NAME,
C_FILE_NAME if wants_images else C_OBJECTS_FILE_NAME,
C_MD5_DIGEST,
C_SCALING,
C_HEIGHT,
C_WIDTH,
)
]
pathname = m.get_current_image_measurement(m_path)
filename = m.get_current_image_measurement(m_file)
rescale = wants_images and file_setting.rescale.value
provider = LoadImagesImageProvider(image_name, pathname, filename,
rescale)
image = provider.provide_image(image_set)
pixel_data = image.pixel_data
if wants_images:
md5 = provider.get_md5_hash(m)
m.add_image_measurement("_".join((C_MD5_DIGEST, image_name)),
md5)
m.add_image_measurement("_".join((C_SCALING, image_name)),
image.scale)
m.add_image_measurement("_".join((C_HEIGHT, image_name)),
int(pixel_data.shape[0]))
m.add_image_measurement("_".join((C_WIDTH, image_name)),
int(pixel_data.shape[1]))
image_set.providers.append(provider)
else:
#
# Turn image into objects
#
labels = convert_image_to_objects(pixel_data)
objects = cpo.Objects()
objects.segmented = labels
object_set = workspace.object_set
assert isinstance(object_set, cpo.ObjectSet)
object_set.add_objects(objects, image_name)
add_object_count_measurements(m, image_name, objects.count)
add_object_location_measurements(m, image_name, labels)
#
# Add outlines if appropriate
#
if file_setting.wants_outlines:
outlines = centrosome.outline.outline(labels)
outline_image = cpi.Image(outlines.astype(bool))
workspace.image_set.add(file_setting.outlines_name.value,
outline_image)
statistics += [(image_name, filename)]
workspace.display_data.col_labels = ("Image name", "File")
workspace.display_data.statistics = statistics
def run(self, workspace):
statistics = []
m = workspace.measurements
assert isinstance(m, cpmeas.Measurements)
#
# Hack: if LoadSingleImage is first, no paths are populated
#
if self.file_wants_images(self.file_settings[0]):
m_path = "_".join((C_PATH_NAME,
self.file_settings[0].image_name.value))
else:
m_path = "_".join((C_OBJECTS_PATH_NAME,
self.file_settings[0].objects_name.value))
if m.get_current_image_measurement(m_path) is None:
self.prepare_run(workspace)
image_set = workspace.image_set
for file_setting in self.file_settings:
wants_images = self.file_wants_images(file_setting)
image_name = file_setting.image_name.value if wants_images else \
file_setting.objects_name.value
m_path, m_file, m_md5_digest, m_scaling, m_height, m_width = [
"_".join((c, image_name)) for c in (
C_PATH_NAME if wants_images else C_OBJECTS_PATH_NAME,
C_FILE_NAME if wants_images else C_OBJECTS_FILE_NAME,
C_MD5_DIGEST, C_SCALING, C_HEIGHT, C_WIDTH)]
pathname = m.get_current_image_measurement(m_path)
filename = m.get_current_image_measurement(m_file)
rescale = (wants_images and file_setting.rescale.value)
provider = LoadImagesImageProvider(
image_name, pathname, filename, rescale)
image = provider.provide_image(image_set)
pixel_data = image.pixel_data
if wants_images:
md5 = provider.get_md5_hash(m)
m.add_image_measurement("_".join((C_MD5_DIGEST, image_name)),
md5)
m.add_image_measurement("_".join((C_SCALING, image_name)),
image.scale)
m.add_image_measurement("_".join((C_HEIGHT, image_name)),
int(pixel_data.shape[0]))
m.add_image_measurement("_".join((C_WIDTH, image_name)),
int(pixel_data.shape[1]))
image_set.providers.append(provider)
else:
#
# Turn image into objects
#
labels = convert_image_to_objects(pixel_data)
objects = cpo.Objects()
objects.segmented = labels
object_set = workspace.object_set
assert isinstance(object_set, cpo.ObjectSet)
object_set.add_objects(objects, image_name)
add_object_count_measurements(m, image_name, objects.count)
add_object_location_measurements(m, image_name, labels)
#
# Add outlines if appropriate
#
if file_setting.wants_outlines:
outlines = centrosome.outline.outline(labels)
outline_image = cpi.Image(outlines.astype(bool))
workspace.image_set.add(file_setting.outlines_name.value,
outline_image)
statistics += [(image_name, filename)]
workspace.display_data.col_labels = ("Image name", "File")
workspace.display_data.statistics = statistics
