def run(self, workspace):
import matplotlib
import matplotlib.cm
import matplotlib.backends.backend_agg
import matplotlib.transforms
from cellprofiler.gui.tools import figure_to_image, only_display_image
#
# Get the image
#
image = workspace.image_set.get_image(self.image_name.value)
if self.wants_image:
pixel_data = image.pixel_data
else:
pixel_data = np.zeros(image.pixel_data.shape[:2])
object_set = workspace.object_set
if self.objects_or_image == OI_OBJECTS:
if self.objects_name.value in object_set.get_object_names():
objects = object_set.get_objects(self.objects_name.value)
else:
objects = None
workspace.display_data.pixel_data = pixel_data
if self.use_color_map():
workspace.display_data.labels = objects.segmented
#
# Get the measurements and positions
#
measurements = workspace.measurements
if self.objects_or_image == OI_IMAGE:
value = measurements.get_current_image_measurement(self.measurement.value)
values = [value]
x = [pixel_data.shape[1] / 2]
x_offset = np.random.uniform(high=1.0, low=-1.0)
x[0] += x_offset
y = [pixel_data.shape[0] / 2]
y_offset = np.sqrt(1 - x_offset ** 2)
y[0] += y_offset
else:
values = measurements.get_current_measurement(self.objects_name.value, self.measurement.value)
if objects is not None and len(values) < objects.count:
temp = np.zeros(objects.count, values.dtype)
temp[: len(values)] = values
temp[len(values) :] = np.nan
values = temp
x = measurements.get_current_measurement(self.objects_name.value, M_LOCATION_CENTER_X)
x_offset = np.random.uniform(high=1.0, low=-1.0, size=x.shape)
y_offset = np.sqrt(1 - x_offset ** 2)
x += self.offset.value * x_offset
y = measurements.get_current_measurement(self.objects_name.value, M_LOCATION_CENTER_Y)
y += self.offset.value * y_offset
mask = ~(np.isnan(values) | np.isnan(x) | np.isnan(y))
values = values[mask]
x = x[mask]
y = y[mask]
workspace.display_data.mask = mask
workspace.display_data.values = values
workspace.display_data.x = x
workspace.display_data.y = y
fig = matplotlib.figure.Figure()
axes = fig.add_subplot(1, 1, 1)
def imshow_fn(pixel_data):
# Note: requires typecast to avoid failure during
# figure_to_image (IMG-764)
img = pixel_data * 255
img[img < 0] = 0
img[img > 255] = 255
img = img.astype(np.uint8)
axes.imshow(img, cmap=matplotlib.cm.Greys_r)
self.display_on_figure(workspace, axes, imshow_fn)
canvas = matplotlib.backends.backend_agg.FigureCanvasAgg(fig)
if self.saved_image_contents == E_AXES:
fig.set_frameon(False)
if not self.use_color_map():
fig.subplots_adjust(0.1, 0.1, 0.9, 0.9, 0, 0)
shape = pixel_data.shape
width = float(shape[1]) / fig.dpi
height = float(shape[0]) / fig.dpi
fig.set_figheight(height)
fig.set_figwidth(width)
elif self.saved_image_contents == E_IMAGE:
if self.use_color_map():
fig.axes[1].set_visible(False)
only_display_image(fig, pixel_data.shape)
else:
if not self.use_color_map():
fig.subplots_adjust(0.1, 0.1, 0.9, 0.9, 0, 0)
pixel_data = figure_to_image(fig, dpi=fig.dpi)
image = cpi.Image(pixel_data)
workspace.image_set.add(self.display_image.value, image)
def run(self, workspace):
import matplotlib
import matplotlib.cm
import matplotlib.backends.backend_agg
import matplotlib.transforms
from cellprofiler.gui.tools import figure_to_image, only_display_image
#
# Get the image
#
image = workspace.image_set.get_image(self.image_name.value)
if self.wants_image:
pixel_data = image.pixel_data
else:
pixel_data = np.zeros(image.pixel_data.shape[:2])
object_set = workspace.object_set
if self.objects_or_image == OI_OBJECTS:
if self.objects_name.value in object_set.get_object_names():
objects = object_set.get_objects(self.objects_name.value)
else:
objects = None
workspace.display_data.pixel_data = pixel_data
if self.use_color_map():
workspace.display_data.labels = objects.segmented
#
# Get the measurements and positions
#
measurements = workspace.measurements
if self.objects_or_image == OI_IMAGE:
value = measurements.get_current_image_measurement(
self.measurement.value)
values = [value]
x = [pixel_data.shape[1] / 2]
x_offset = np.random.uniform(high=1.0, low=-1.0)
x[0] += x_offset
y = [pixel_data.shape[0] / 2]
y_offset = np.sqrt(1 - x_offset**2)
y[0] += y_offset
else:
values = measurements.get_current_measurement(
self.objects_name.value, self.measurement.value)
if objects is not None:
if len(values) < objects.count:
temp = np.zeros(objects.count, values.dtype)
temp[:len(values)] = values
temp[len(values):] = np.nan
values = temp
elif len(values) > objects.count:
# If the values for something (say, object number) are greater
# than the actual number of objects we have, some might have been
# filtered out/removed. We'll need to diff the arrays to figure out
# what objects to remove
indices = objects.indices
diff = np.setdiff1d(indices, np.unique(objects.segmented))
values = np.delete(values, diff)
x = measurements.get_current_measurement(self.objects_name.value,
M_LOCATION_CENTER_X)
x_offset = np.random.uniform(high=1.0, low=-1.0, size=x.shape)
y_offset = np.sqrt(1 - x_offset**2)
x += self.offset.value * x_offset
y = measurements.get_current_measurement(self.objects_name.value,
M_LOCATION_CENTER_Y)
y += self.offset.value * y_offset
mask = ~(np.isnan(values) | np.isnan(x) | np.isnan(y))
values = values[mask]
x = x[mask]
y = y[mask]
workspace.display_data.mask = mask
workspace.display_data.values = values
workspace.display_data.x = x
workspace.display_data.y = y
fig = matplotlib.figure.Figure()
axes = fig.add_subplot(1, 1, 1)
def imshow_fn(pixel_data):
# Note: requires typecast to avoid failure during
# figure_to_image (IMG-764)
img = pixel_data * 255
img[img < 0] = 0
img[img > 255] = 255
img = img.astype(np.uint8)
axes.imshow(img, cmap=matplotlib.cm.Greys_r)
self.display_on_figure(workspace, axes, imshow_fn)
canvas = matplotlib.backends.backend_agg.FigureCanvasAgg(fig)
if self.saved_image_contents == E_AXES:
fig.set_frameon(False)
if not self.use_color_map():
fig.subplots_adjust(0.1, .1, .9, .9, 0, 0)
shape = pixel_data.shape
width = float(shape[1]) / fig.dpi
height = float(shape[0]) / fig.dpi
fig.set_figheight(height)
fig.set_figwidth(width)
elif self.saved_image_contents == E_IMAGE:
if self.use_color_map():
fig.axes[1].set_visible(False)
only_display_image(fig, pixel_data.shape)
else:
if not self.use_color_map():
fig.subplots_adjust(.1, .1, .9, .9, 0, 0)
pixel_data = figure_to_image(fig, dpi=fig.dpi)
image = cpi.Image(pixel_data)
workspace.image_set.add(self.display_image.value, image)
