toupy.registration package

Submodules

toupy.registration.registration module

toupy.registration.registration.alignprojections_horizontal(sinogram, theta, shiftstack, **params)[source]

Function to align projections by tomographic consistency 1, 2. It relies on having already aligned the vertical direction. The code aligns using the consistency before and after tomographic combination of projections.

Parameters

  • sinogram (array_like) – Sinogram derivative, the second index should be the angle

  • theta (array_like) – Reconstruction angles (in degrees). Default: m angles evenly spaced between 0 and 180 (if the shape of radon_image is (N, M)).

  • shiftstack (array_like) – Array with initial estimates of positions

  • params (dict) – Container with parameters for the registration

  • params["pixtol"] (float) – Tolerance for alignment, which is also used as a search step

  • params["alignx"] (bool) – True or False to activate align x using center of mass (default= False, which means align y only)

  • params["shiftmeth"] (str) – Shift images with fourier method (default). The options are linear -> Shift images with linear interpolation (default); fourier -> Fourier shift or spline -> Shift images with spline interpolation.

  • params["circle"] (bool) – Use a circular mask to eliminate corners of the tomogram

  • params["filtertype"] (str) – Filter to use for FBP

  • params["freqcutoff"] (float) – Frequency cutoff for tomography filter (between 0 and 1)

  • params["cliplow"] (float) – Minimum value in tomogram

  • params["cliphigh"] (float) – Maximum value in tomogram

Returns

  • shiftstack (array_like) – Corrected object positions

  • alinedsinogram (array_like) – Array containting the aligned sinogram

References

1

Guizar-Sicairos, M., et al., “Quantitative interior x-ray nanotomography by a hybrid imaging technique,” Optica 2, 259-266 (2015).

2

da Silva, J. C., et al., “High energy near-and far-field ptychographic tomography at the ESRF,” Proc. SPIE 10391, Developments in X-Ray Tomography XI, 1039106 (2017).

toupy.registration.registration.alignprojections_vertical(input_stack, shiftstack, **params)[source]

Vertical alignment of projections using mass fluctuation approach 3, 4. It relies on having air on both sides of the sample (non local tomography). It performs a local search in y, so convergence issues can be addressed by giving an approximate initial guess for a possible drift via shiftstack

Parameters

  • input_stack (array_like) – Stack of projections

  • limrow (list of ints) – Limits of window of interest in y

  • limcol (list of ints) – Limits of window of interest in x

  • shiftstack (array_like) – Array of initial estimates for object motion (2,n)

  • params (dict) – Container with parameters for the registration

  • params['pixtol'] (float) – Tolerance for alignment, which is also used as a search step

  • params['polyorder'] (int) – Specify the polynomial order of bias removal. For example: polyorder = 1 -> mean, polyorder = 2 -> linear).

  • params['alignx'] (bool) – True or False to activate align x using center of mass (default= False, which means align y only)

  • params['shiftmeth'] (str) – Shift images with fourier method (default). The options are linear -> Shift images with linear interpolation (default); fourier -> Fourier shift or spline -> Shift images with spline interpolation.

Returns

  • shiftstack (array_like) – Corrected bject positions

  • input_stack (array_like) – Aligned stack of the projections

References

3

Guizar-Sicairos, M., et al. , “Phase tomography from x-ray coherent diffractive imaging projections,” Opt. Express 19, 21345-21357 (2011).

4

da Silva, J. C., et al. “High energy near-and far-field ptychographic tomography at the ESRF,” Proc. SPIE 10391, Developments in X-Ray Tomography XI, 1039106 (2017)

toupy.registration.registration.center_of_mass_stack(input_stack, lims, shiftstack, shift_method='fourier')[source]

Calculates the center of the mass for each projection in the stack and returns a stack of centers of mass (row, col) i.e., returns shiftstack[1] If the array is zero, it return the center of mass at 0.

toupy.registration.registration.compute_aligned_horizontal(input_stack, shiftstack, shift_method='linear')[source]

Compute the alignment of the stack on at the horizontal direction

Parameters

  • input_array (array_like) – Stack of images to be shifted

  • shiftstack (array_like) – Array of initial estimates for object motion (2,n) The estimates for vertical movement will be changed to 0

  • shift_method (str (default linear)) – Name of the shift method. Options: ‘linear’, ‘fourier’, ‘spline’

Returns

output_stack – 2D function containing the stack of aligned images

Return type

array_like

toupy.registration.registration.compute_aligned_sino(input_sino, shiftslice, shift_method='linear')[source]

Compute the aligned sinogram given the correction for object positions

Parameters

  • input_sino (array_like) – Input sinogram to be shifted

  • shiftslice (array_like) – Array of estimates for object motion (1,n)

  • shift_method (str (default linear)) – Name of the shift method. Options: ‘linear’, ‘fourier’, ‘spline’

Returns

output_sino – 2D function containing the aligned sinogram

Return type

array_like

toupy.registration.registration.compute_aligned_stack(input_stack, shiftstack, shift_method='linear')[source]

Compute the aligned stack given the correction for object positions

Parameters

  • input_array (array_like) – Stack of images to be shifted

  • shiftstack (array_like) – Array of initial estimates for object motion (2,n)

  • shift_method (str (default linear)) – Name of the shift method. Options: ‘linear’, ‘fourier’, ‘spline’

Returns

output_stack – 2D function containing the stack of aligned images

Return type

array_like

toupy.registration.registration.estimate_rot_axis(input_array, theta, **params)[source]

Initial estimate of the rotation axis

toupy.registration.registration.oneslicefordisplay(sinogram, theta, **params)[source]

Calculate one slice for display.

Parameters

  • sinogram (array_like) – Sinogram derivative, the second index should be the angle

  • theta (array_like) – Reconstruction angles (in degrees). Default: m angles evenly spaced between 0 and 180 (if the shape of radon_image is (N, M)).

  • params (dict) – Container with parameters for the registration.

  • params["filtertype"] (str) – Filter to use for FBP

  • params["freqcutoff"] (float) – Frequency cutoff for tomography filter (between 0 and 1)

toupy.registration.registration.refine_horizontalalignment(input_stack, theta, shiftstack, **params)[source]

Refine horizontal alignment. Please, see the description of each parameter in alignprojections_horizontal().

toupy.registration.registration.register_2Darrays(image1, image2)[source]

Image registration. Register two images using phase cross correlations.

Parameters

  • image1 (array_like) – Image of reference

  • image2 (array_like) – Image to be shifted relative to image1

Returns

  • shift (list of floats) – List of shifts applied, with the row shift in the 1st dimension and the column shift in the 2nd dimension.

  • diffphase (float) – Difference of phase between the two images

  • offset_image2 (array_like) – Shifted image2 relative to image1

toupy.registration.registration.tomoconsistency_multiple(input_stack, theta, shiftstack, **params)[source]

Apply tomographic consistency alignement on multiple slices. By default is implemented over 10 slices.

Parameters

  • Input_stack (array_like) – Stack of projections

  • theta (array_like) – Reconstruction angles (in degrees). Default: m angles evenly spaced between 0 and 180 (if the shape of radon_image is (N, M)).

  • shiftstack (array_like) – Array with initial estimates of positions

  • params (dict) – Dictionary with additional parameters for the alignment. Please, see the description of each parameter in alignprojections_horizontal().

Returns

shiftstack – Average of the object shifts over 10 slices

Return type

array_like

toupy.registration.registration.vertical_fluctuations(input_stack, lims, shiftstack, shift_method='fourier', polyorder=2)[source]

Calculate the vertical fluctuation functions of a stack

Parameters

  • input_array (array_like) – Stack of images to be shifted

  • lims (list of ints) – Limits of rows and columns to be considered. lims=[limrow,limcol]

  • shiftstack (array_like) – Array of initial estimates for object motion (2,n)

  • shift_method (str, optional) – Name of the shift method. Options: ‘linear’, ‘fourier’, ‘spline’. The default method is ‘linear’.

  • polyorder (int, optional) – Order of the polynomial to remove bias from the mass fluctuation function. The default value is 2.

Returns

vert_fluct – 2D function containing the mass fluctuation after shift and bias removal for the stack of images

Return type

array_like

toupy.registration.registration.vertical_shift(input_array, lims, vstep, maxshift, shift_method='linear', polyorder=2)[source]

Calculate the vertical shift of an array

Parameters

  • input_array (array_like) – Image to be shifted

  • lims (list of ints) – Limits of rows and columns to be considered. lims=[limrow,limcol]

  • vstep (float) – Amount to shift the input_array vertically

  • maxshift (float) – Maximum value of the shifts in order to avoid border problems

  • shift_method (str, optional) – Name of the shift method. Options: ‘linear’, ‘fourier’, ‘spline’. The default method is ‘linear’.

  • polyorder (int, optional) – Order of the polynomial to remove bias from the mass fluctuation function. The default value is 2.

Returns

shift_cal – 1D function containing the mass fluctuation after shift and bias removal

Return type

array_like

toupy.registration.shift module

class toupy.registration.shift.ShiftFunc(**params)[source]

Bases: Variables

Collections of shift fuctions

__call__(*args)[source]

Implement the shifts

Parameters

*args –

args[0]array_like

Input array

args[1]int or tuple

Shift amplitude

args[2]str (optional)

Padding mode if necessary

args[3]bool (optional)

True for complex output or False for real output

shift_fft(input_array, shift)[source]

Performs pixel and subpixel shift (with wraping) using pyFFTW.

Since FFTW has efficient functions for array sizes which can be decompose in prime factor, the input_array is padded to the next fast size given by pyFFTW.next_fast_len. The padding is done in mode = ‘reflect’ by default to reduce border artifacts.

Parameters

  • input_array (array_like) – Input image to calculate the shifts.

  • shift (int or tuple) – Number of pixels to shift. For 1D, use a integer value. For 2D, use a tuple of integers where the first value corresponds to shifts in the rows and the second value corresponds to shifts in the columns.

Returns

output_array – Shifted image

Return type

array_like

shift_linear(input_array, shift)[source]

Shifts an image with wrap around and bilinear interpolation

Parameters

  • input_array (array_like) – Input image to calculate the shifts.

  • shift (int or tuple) – Number of pixels to shift. For 1D, use a integer value. For 2D, use a tuple of integers where the first value corresponds to shifts in the rows and the second value corresponds to shifts in the columns.

Returns

output_array – Shifted image

Return type

array_like

shift_spline_wrap(input_array, shift)[source]

Performs pixel and subpixel shift (with wraping) using splines

Parameters

  • input_array (array_like) – Input image to calculate the shifts.

  • shift (int or tuple) – Number of pixels to shift. For 1D, use a integer value. For 2D, use a tuple of integers where the first value corresponds to shifts in the rows and the second value corresponds to shifts in the columns.

Returns

output_array – Shifted image

Return type

array_like