def get_background_model(self, saturation_radius):
"""Creates a model for the background of the signal.
The mean radial profile is fitted with the following three components:
* Voigt profile for the central beam
* Exponential profile for the diffuse scatter
* Linear profile for the background offset and to improve the fit
Using the exponential profile and the linear profile, an
ElectronDiffraction signal is produced representing the mean background
of the signal. This may be used for background subtraction.
Parameters
----------
saturation_radius : int
The radius of the region about the central beam in which pixels are
saturated.
Returns
-------
ElectronDiffraction
The mean background of the signal.
"""
# TODO: get this done without taking the mean
profile = self.get_radial_profile().mean()
model = profile.create_model()
e1 = saturation_radius * profile.axes_manager.signal_axes[0].scale
model.set_signal_range(e1)
direct_beam = Voigt()
direct_beam.centre.value = 0
direct_beam.centre.free = False
direct_beam.FWHM.value = 0.1
direct_beam.area.bmin = 0
model.append(direct_beam)
diffuse_scatter = Exponential()
diffuse_scatter.A.value = 0
diffuse_scatter.A.bmin = 0
diffuse_scatter.tau.value = 0
diffuse_scatter.tau.bmin = 0
model.append(diffuse_scatter)
linear_decay = Polynomial(1)
model.append(linear_decay)
model.fit(bounded=True)
x_axis = self.axes_manager.signal_axes[0].axis
y_axis = self.axes_manager.signal_axes[1].axis
xs, ys = np.meshgrid(x_axis, y_axis)
rs = (xs ** 2 + ys ** 2) ** 0.5
bg = ElectronDiffraction(
diffuse_scatter.function(rs) + linear_decay.function(rs))
for i in (0, 1):
bg.axes_manager.signal_axes[i].update_from(
self.axes_manager.signal_axes[i])
return bg
def _make_radius_vs_angle_model(
signal,
radial_signal_span,
angleN=15,
centre_x=None,
centre_y=None,
prepeak_range=None,
show_progressbar=True,
):
s_ra = signal.angular_slice_radial_average(
angleN=angleN,
centre_x=centre_x,
centre_y=centre_y,
show_progressbar=show_progressbar,
)
s_ra = s_ra.isig[radial_signal_span[0]:radial_signal_span[1]]
m_ra = s_ra.create_model()
# Fit 1 order polynomial to edges of data to account for the background
sa = m_ra.axes_manager.signal_axes[0]
if prepeak_range is None:
prepeak_range = (sa.low_value, sa.index2value(4))
m_ra.set_signal_range(prepeak_range[0], prepeak_range[1])
m_ra.add_signal_range(sa.index2value(-3), sa.high_value)
polynomial = Polynomial(1)
m_ra.append(polynomial)
m_ra.multifit(show_progressbar=show_progressbar)
polynomial.set_parameters_not_free()
m_ra.reset_signal_range()
# Fit Gaussian to diffraction ring
centre_initial = (sa.high_value + sa.low_value) * 0.5
sigma = 3
A = (s_ra - s_ra.min(axis=1)).data.max() * 2 * sigma
gaussian = Gaussian(A=A, sigma=sigma, centre=centre_initial)
gaussian.centre.bmin = sa.low_value
gaussian.centre.bmax = sa.high_value
m_ra.append(gaussian)
return m_ra