def make_background(self):
import copy
if hasattr(self, "background"):
return self.background
# raw background data
background = copy.deepcopy(self.raw_data).as_double()
# mask out the signal areas
mask = self.get_signal_mask()
background.as_1d().set_selected(mask.as_1d(), 0.0)
inv_mask = (~mask).as_1d().as_int()
inv_mask.reshape(self.raw_data.accessor())
from dials.algorithms.image.filter import summed_area
from dials.array_family import flex
summed_background = summed_area(background, (5, 5))
summed_mask = summed_area(inv_mask, (5, 5))
mean_background = summed_background / summed_mask.as_double()
background.as_1d().set_selected(mask.as_1d(), mean_background.as_1d())
self.background = background
return background
def fit(data, mask, model_data, model_mask, kernel):
from dials.array_family import flex
from dials.algorithms.image.filter import summed_area
mask = mask & model_mask
mask = mask.as_1d().as_int()
mask.reshape(data.accessor())
data = data * mask
model_data = model_data * mask.as_double()
summed_data = summed_area(data, kernel).as_1d()
summed_mask = summed_area(mask, kernel).as_1d()
summed_model = summed_area(model_data, kernel).as_1d()
scaled_mask = (model_mask.as_1d() == True) & (summed_mask > 2) & (summed_model > 0)
indices = flex.size_t(range(len(data))).select(scaled_mask)
scaled_data = flex.double(len(data))
scaled_data.set_selected(
indices,
summed_data.select(indices).as_double() / summed_model.select(indices))
scaled_data.reshape(data.accessor())
scaled_mask.reshape(data.accessor())
#for j in range(scaled_data.all()[0]):
# last_scale = None
# for i in range(scaled_data.all()[1]):
# if scaled_mask[j,i] == False:
# if last_scale != None:
# scaled_data[j,i] = last_scale
# scaled_mask[j,i] = True
# else:
# last_scale = scaled_data[j,i]
return scaled_data, scaled_mask
def fit(data, mask, model_data, model_mask, kernel):
from dials.array_family import flex
from dials.algorithms.image.filter import summed_area
mask = mask & model_mask
mask = mask.as_1d().as_int()
mask.reshape(data.accessor())
data = data * mask
model_data = model_data * mask.as_double()
summed_data = summed_area(data, kernel).as_1d()
summed_mask = summed_area(mask, kernel).as_1d()
summed_model = summed_area(model_data, kernel).as_1d()
scaled_mask = (model_mask.as_1d()
== True) & (summed_mask > 2) & (summed_model > 0)
indices = flex.size_t(range(len(data))).select(scaled_mask)
scaled_data = flex.double(len(data))
scaled_data.set_selected(
indices,
summed_data.select(indices).as_double() / summed_model.select(indices))
scaled_data.reshape(data.accessor())
scaled_mask.reshape(data.accessor())
# for j in range(scaled_data.all()[0]):
# last_scale = None
# for i in range(scaled_data.all()[1]):
# if scaled_mask[j,i] == False:
# if last_scale != None:
# scaled_data[j,i] = last_scale
# scaled_mask[j,i] = True
# else:
# last_scale = scaled_data[j,i]
return scaled_data, scaled_mask
def test():
from dials.algorithms.image.filter import summed_area
from scitbx.array_family import flex
# Create an image
image = flex.random_double(2000 * 2000)
image.reshape(flex.grid(2000, 2000))
# Calculate the summed area table
sa = summed_area(image, (3, 3))
# For a selection of random points, ensure that the value is the
# sum of the area under the kernel
for i in range(10000):
i = random.randint(10, 1990)
j = random.randint(10, 1990)
v = sa[j, i]
e = flex.sum(image[j - 3:j + 4, i - 3:i + 4])
assert e == pytest.approx(v, abs=1e-7)
def run(self):
from dials.algorithms.image.filter import summed_area
from scitbx.array_family import flex
from random import randint
# Create an image
image = flex.random_double(2000 * 2000)
image.reshape(flex.grid(2000, 2000))
# Calculate the summed area table
sa = summed_area(image, (3, 3))
# For a selection of random points, ensure that the value is the
# sum of the area under the kernel
eps = 1e-7
for i in range(10000):
i = randint(10, 1990)
j = randint(10, 1990)
v = sa[j, i]
e = flex.sum(image[j - 3:j + 4, i - 3:i + 4])
assert (abs(e - v) <= eps)
# Test passed
print 'OK'
def run(self):
"""Extract the shoeboxes."""
from dials.util.options import flatten_reflections
from dials.util.options import flatten_experiments
from dials.util.options import flatten_experiments
from dials.util import log
from dials.array_family import flex
from dials.util import Sorry
# Parse the command line
params, options = self.parser.parse_args(show_diff_phil=False)
# Configure logging
log.config()
# Log the diff phil
diff_phil = self.parser.diff_phil.as_str()
if diff_phil is not "":
logger.info("The following parameters have been modified:\n")
logger.info(diff_phil)
# Get the data
reflections = flatten_reflections(params.input.reflections)
experiments = flatten_experiments(params.input.experiments)
if not any([experiments, reflections]):
self.parser.print_help()
exit(0)
elif len(experiments) > 1:
raise Sorry("More than 1 experiment set")
elif len(experiments) == 1:
imageset = experiments[0].imageset
if len(reflections) != 1:
raise Sorry("Need 1 reflection table, got %d" % len(reflections))
else:
reflections = reflections[0]
# Check the reflections contain the necessary stuff
assert "bbox" in reflections
assert "panel" in reflections
# Get some models
detector = imageset.get_detector()
scan = imageset.get_scan()
frame0, frame1 = scan.get_array_range()
# Add some padding but limit to image volume
if params.padding > 0:
logger.info("Adding %d pixels as padding" % params.padding)
x0, x1, y0, y1, z0, z1 = reflections["bbox"].parts()
x0 -= params.padding
x1 += params.padding
y0 -= params.padding
y1 += params.padding
panel = reflections["panel"]
for i in range(len(reflections)):
width, height = detector[panel[i]].get_image_size()
if x0[i] < 0:
x0[i] = 0
if x1[i] > width:
x1[i] = width
if y0[i] < 0:
y0[i] = 0
if y1[i] > height:
y1[i] = height
reflections["bbox"] = flex.int6(x0, x1, y0, y1, z0, z1)
# Now iterate through the images, masking the shoebox pixels as necessary,
# then model the background & compute intensity statistics
x0, x1, y0, y1, z0, z1 = reflections["bbox"].parts()
for frame in range(frame0, frame1):
data = imageset.get_raw_data(frame)[0]
mask = data < 0
data = data.as_double()
sel = (z0 <= frame) & (z1 >= frame)
_x0 = x0.select(sel)
_x1 = x1.select(sel)
_y0 = y0.select(sel)
_y1 = y1.select(sel)
for __x0, __x1, __y0, __y1 in zip(_x0, _x1, _y0, _y1):
for y in range(__y0, __y1):
for x in range(__x0, __x1):
data[y, x] = 0.0
mask[y, x] = True
imask = (~mask).as_1d().as_int()
imask.reshape(data.accessor())
from dials.algorithms.image.filter import summed_area
# smooth over scale corresponding to max shoebox size
d = max(flex.max(_x1 - _x0), flex.max(_y1 - _y0))
summed_background = summed_area(data, (d, d))
summed_mask = summed_area(imask, (d, d))
mean_background = summed_background / summed_mask.as_double()
data.as_1d().set_selected(mask.as_1d(), mean_background.as_1d())
print(flex.sum(data.select(data.as_1d() > 0)))
