from typing import Tuple, cast
from numpy import (
broadcast_to,
cos,
divide,
dtype,
expand_dims,
floating,
log,
logical_and,
ndarray,
number,
ones_like,
power,
)
from ..utils.geometry import scattering_angles
from ..utils.typing import FrameHeight, Frames, FrameWidth, NumFrames, VectorOrSingle
[docs]
def correct_self_absorption(
frames: Frames[NumFrames, FrameWidth, FrameHeight, dtype[number]],
incident_flux: ndarray[VectorOrSingle[NumFrames], dtype[number]],
transmitted_flux: ndarray[VectorOrSingle[NumFrames], dtype[number]],
beam_center: Tuple[float, float],
pixel_sizes: Tuple[float, float],
distance: float,
) -> Frames[NumFrames, FrameWidth, FrameHeight, dtype[number]]:
"""Correct for transmission loss due to differences in observation angle.
Correct for transmission loss due to differences in observation angle, as detailed
in section 3.4.7 of 'Everything SAXS: small-angle scattering pattern collection and
correction' [https://doi.org/10.1088/0953-8984/25/38/383201].
Args:
frames: A stack of frames to be corrected.
incident_flux: The flux intensity observed upstream of the sample.
transmitted_flux: The flux intensity observed downstream of the sample.
beam_center: The center position of the beam in pixels.
pixel_sizes: The real space size of a detector pixel.
distance: The distance between the detector and the sample.
Returns:
The corrected stack of frames.
"""
angles = scattering_angles(
cast(Tuple[int, int], frames.shape[-2:]), beam_center, pixel_sizes, distance
)
transmissibility = transmitted_flux / incident_flux
transmissibility = (
transmissibility
if transmissibility.shape == (1,)
else expand_dims(transmissibility, (1, 2))
)
secangles: ndarray[Tuple[int, int], dtype[floating]] = broadcast_to(
1 / cos(angles), frames.shape
)
correction_factors: ndarray[Tuple[int, int, int], dtype[floating]] = divide(
1 - power(transmissibility, secangles - 1),
log(transmissibility) * (1 - secangles),
out=ones_like(secangles),
where=logical_and(secangles != 1.0, transmissibility != 1.0),
)
return frames * correction_factors
