corrct.models

Define all the models used through-out the code.

@author: Nicola VIGANÒ, Computational Imaging group, CWI, The Netherlands, and ESRF - The European Synchrotron, Grenoble, France

Module Contents

Classes

Geometry

Base geometry class.

ProjectionGeometry

Store the projection geometry.

VolumeGeometry

Store the volume geometry.

Functions

combine_shifts_vu

Combine vertical and horizontal shifts.

get_rot_axis_dir

Process the requested rotation axis direction and return a meaningful value.

_get_data_dims

get_prj_geom_parallel

Generate the default geometry for parallel beam.

get_prj_geom_cone

Generate the default geometry for parallel beam.

get_vol_geom_from_data

Generate a default volume geometry from the data shape.

get_vol_geom_from_volume

Generate a default volume geometry from the given volume.

Data

ROT_DIRS_VALID

API

corrct.models.ROT_DIRS_VALID

(‘clockwise’, ‘counter-clockwise’)

class corrct.models.Geometry[source]

Bases: abc.ABC

Base geometry class.

__str__() → str[source]

Return a human readable representation of the object.

Returns

str The human readable representation of the object.

class corrct.models.ProjectionGeometry[source]

Bases: corrct.models.Geometry

Store the projection geometry.

geom_type: str

None

src_pos_xyz: numpy.typing.NDArray

None

det_pos_xyz: numpy.typing.NDArray

None

det_u_xyz: numpy.typing.NDArray

None

det_v_xyz: numpy.typing.NDArray

None

rot_dir_xyz: numpy.typing.NDArray

None

pix2vox_ratio: float

1

det_shape_vu: Optional[numpy.typing.NDArray]

None

__post_init__() → None[source]
__getitem__(indx: Any)[source]

Slice projection geometry along the angular direction.

Parameters

indx : Any Indices of the slicing.

copy() → corrct.models.ProjectionGeometry[source]

Deepcopy an existing geometry.

Returns

ProjectionGeometry The new instance of ProjectionGeometry

static get_default_parallel(*, geom_type: str = '3d', rot_axis_shift_pix: Optional[numpy.typing.ArrayLike] = None, rot_axis_dir: Union[str, numpy.typing.ArrayLike] = 'clockwise') → corrct.models.ProjectionGeometry[source]

Generate the default geometry for parallel beam.

Parameters

geom_type : str, optional The geometry type. The default is “parallel3d”. rot_axis_shift_pix : Optional[ArrayLike], optional Rotation axis shift in pixels. The default is None. rot_axis_dir : Union[str, ArrayLike], optional Rotation axis direction. It can be either a string or a direction. The default is “clockwise”.

Returns

ProjectionGeometry The default parallel-beam geometry.

property ndim: int

Return the number of dimensions of the geometry.

Returns

int The number of dimensions.

get_3d() → corrct.models.ProjectionGeometry[source]

Return the 3D version of the geometry.

Returns

ProjectionGeometry The new geometry.

set_detector_shape_vu(vu: Union[int, Sequence[int], numpy.typing.NDArray]) → None[source]

Set the detector VU shape.

Parameters

vu : int | Sequence[int] | NDArray The VU shape of the projection data.

set_detector_shifts_vu(det_pos_vu: Union[numpy.typing.ArrayLike, numpy.typing.NDArray, None] = None, cor_pos_u: Union[float, None] = None, det_dist_y: numpy.typing.ArrayLike = 0.0) → None[source]

Set the detector position in XZ, from VU (vertical, horizontal) coordinates.

Parameters

det_pos_vu : ArrayLike | NDArray | None Detector vertical and horizontal positions. Vertical is optional. cor_pos_u : float | None Center of rotation position along U. det_dist_y : ArrayLike, optional Detector distance from origin along Y. The default is 0.0.

set_source_shifts_vu(src_pos_vu: Union[numpy.typing.ArrayLike, numpy.typing.NDArray, None] = None) → None[source]

Set the source position in XZ, from VU (vertical, horizontal) coordinates.

Parameters

src_pos_vu : ArrayLike | NDArray | None Source vertical and horizontal positions. Vertical is optional.

set_detector_tilt(angles_t_rad: Union[numpy.typing.ArrayLike, numpy.typing.NDArray], tilt_axis: Union[Sequence[float], numpy.typing.NDArray] = (0, 1, 0), tilt_source: bool = False) → None[source]

Rotate the detector by the given angle(s) and axis(axes).

Parameters

angles_t_rad : ArrayLike | NDArray Rotation angle(s) in radians. tilt_axis : Sequence[float] | NDArray, optional The tilt axis or axes. The default is (0, 1, 0) tilt_source : bool, optional Whether to also tilt the source. The default is False.

Notes

When applying multiple axes, they will be applied in order. This means that the application is not going to be independent.

rotate(angles_w_rad: numpy.typing.ArrayLike, patch_astra_2d: bool = False) → corrct.models.ProjectionGeometry[source]

Rotate the geometry by the given angle(s).

Parameters

angles_w_rad : ArrayLike Rotation angle(s) in radians.

Returns

ProjectionGeometry The rotated geometry.

get_field_scaled(field_name: str) → numpy.typing.NDArray[source]

Return the a field content, scaled by the pix2vox ratio.

Parameters

field_name : str Name of the field to access.

Returns

NDArray The scaled field.

project_displacement_to_detector(disp_zyx: numpy.typing.ArrayLike) → numpy.typing.NDArray[source]

Project a given displacement vector in the volume coordinates, over the detector.

Parameters

disp_zyx : ArrayLike The displacement vector in volume coordinates.

Returns

NDArray The projection on u (and if applicable v) coordinates.

Raises

ValueError When projection geometry and vector dimensions don match.

get_pre_weights(det_shape_vu: Union[Sequence[int], numpy.typing.NDArray, None] = None) → Union[numpy.typing.NDArray, None][source]

Compute the pre-weights of the projector geometry (notably for cone-beam geometries).

Parameters

det_shape_vu : Sequence[int] | NDArray | None, optional Shape of the detector in [V]U coordinates, by default None

Returns

NDArray | None The computed detector weights

class corrct.models.VolumeGeometry[source]

Bases: corrct.models.Geometry

Store the volume geometry.

_vol_shape_xyz: numpy.typing.NDArray

None

vox_size: float

1.0

__post_init__()[source]

Initialize the input parameters.

is_square() → bool[source]

Compute whether the volume is square in XY.

Returns

bool True is the volume is square in XY.

property shape_xyz: numpy.typing.NDArray

Return the volume shape (XYZ).

Returns

NDArray Shape of the volume (XYZ).

property shape_zxy: numpy.typing.NDArray

Return the volume shape (ZXY).

The swap between X and Y is imposed by the astra-toolbox.

Returns

NDArray Shape of the volume (ZXY).

property mask_shape: numpy.typing.NDArray

Return the XY volume shape for circular masks.

Returns

NDArray Shape of the XY volume.

property extent: Sequence[float]

Return extent of the volume.

Returns

Sequence[float] The extent of the volume [-x, +x, -y, +y, [-z, +z]].

is_3D() → bool[source]

Tell whether this is a 3D geometry.

Returns

bool Whether this is a 3D geometry or not.

get_3d() → corrct.models.VolumeGeometry[source]

Return the 3D version of the geometry.

Returns

VolumeGeometry The new geometry.

static get_default_from_data(data: numpy.typing.NDArray, data_format: str = 'dvwu') → corrct.models.VolumeGeometry[source]

Generate a default volume geometry from the data shape.

Parameters

data : NDArray The data. data_format : str, optional The ordering and meaning of the dimensions in the data. The deault is “dvwu”.

Returns

VolumeGeometry The default volume geometry.

static get_default_from_volume(volume: numpy.typing.NDArray) → corrct.models.VolumeGeometry[source]

Generate a default volume geometry from the given volume.

Parameters

volume : NDArray The volume.

Returns

VolumeGeometry The default volume geometry.

corrct.models.combine_shifts_vu(shifts_v: numpy.typing.NDArray, shifts_u: numpy.typing.NDArray) → numpy.typing.NDArray[source]

Combine vertical and horizontal shifts.

Parameters

shifts_v : NDArray The vertical shifts shifts_u : NDArray The horizontal shifts

Returns

NDArray The combined shifts

corrct.models.get_rot_axis_dir(rot_axis_dir: Union[str, numpy.typing.ArrayLike, numpy.typing.NDArray] = 'clockwise') → numpy.typing.NDArray[source]

Process the requested rotation axis direction and return a meaningful value.

Parameters

rot_axis_dir : Union[str, ArrayLike, NDArray], optional The requested direction, by default “clockwise”

Returns

NDArray The vector corresponding to the rotation direction.

Raises

ValueError In case of malformed direction.

corrct.models._get_data_dims(data_shape: Union[Sequence[int], numpy.typing.NDArray], data_format: str = 'dvwu') → dict[str, Union[int, None]][source]
corrct.models.get_prj_geom_parallel(*, geom_type: str = '3d', rot_axis_shift_pix: Union[numpy.typing.ArrayLike, numpy.typing.NDArray, None] = None, rot_axis_dir: Union[str, numpy.typing.ArrayLike, numpy.typing.NDArray] = 'clockwise', data_shape: Union[Sequence[int], numpy.typing.NDArray, None] = None, data_format: str = 'dvwu') → corrct.models.ProjectionGeometry[source]

Generate the default geometry for parallel beam.

Parameters

geom_type : str, optional The geometry type. The default is “parallel3d”. rot_axis_shift_pix : ArrayLike | NDArray | None, optional Rotation axis shift in pixels. The default is None. rot_axis_dir : str | ArrayLike | NDArray, optional Rotation axis direction. It can be either a string or a direction. The default is “clockwise”.

Returns

ProjectionGeometry The default parallel-beam geometry.

corrct.models.get_prj_geom_cone(*, src_to_sam_dist: float, rot_axis_shift_pix: Union[numpy.typing.ArrayLike, numpy.typing.NDArray, None] = None, rot_axis_dir: Union[str, numpy.typing.ArrayLike, numpy.typing.NDArray] = 'clockwise', data_shape: Union[Sequence[int], numpy.typing.NDArray, None] = None, data_format: str = 'dvwu') → corrct.models.ProjectionGeometry[source]

Generate the default geometry for parallel beam.

Parameters

geom_type : str, optional The geometry type. The default is “parallel3d”. rot_axis_shift_pix : ArrayLike | NDArray | None, optional Rotation axis shift in pixels. The default is None. rot_axis_dir : str | ArrayLike | NDArray, optional Rotation axis direction. It can be either a string or a direction. The default is “clockwise”.

Returns

ProjectionGeometry The default cone-beam geometry.

corrct.models.get_vol_geom_from_data(data: numpy.typing.NDArray, padding_u: Union[int, Sequence[int], numpy.typing.NDArray] = 0, data_format: str = 'dvwu', super_sampling: int = 1) → corrct.models.VolumeGeometry[source]

Generate a default volume geometry from the data shape.

Parameters

data : NDArray The data. padding_u : int | Sequence[int] data_format : str, optional The ordering and meaning of the dimensions in the data. The default is “dvwu”. super_sampling: int, optional The super-sampling size of the voxels. The default is 1.

Returns

VolumeGeometry The default volume geometry.

corrct.models.get_vol_geom_from_volume(volume: numpy.typing.NDArray) → corrct.models.VolumeGeometry[source]

Generate a default volume geometry from the given volume.

Parameters

volume : NDArray The volume.

Returns

VolumeGeometry The default volume geometry.