def get_spectra(self,
streakspeed_in_meV_per_fs,
keep_originals=False,
discretized=True):
'''returns streaked spectra, measured with "number_electronsx" simulated electrons or nondiscretized as a tuple'''
from streaking_cal.statistics import weighted_avg_and_std
if not (self.is_low_res()):
(streaked1, streaked2
) = self.__get_streaked_spectra(streakspeed_in_meV_per_fs)
streaked1 = interp(tof_ens_gpu, self.__eAxis, streaked1)
streaked2 = interp(tof_ens_gpu, self.__eAxis, streaked2)
xuvonly = interp(tof_ens_gpu, self.__eAxis,
cp.square(cp.abs(self.__spec)))
if not (keep_originals):
self.__eAxis = None
self.__spec = None
t_square = cp.square(cp.abs(self.__temp))
t_mean, _ = weighted_avg_and_std(self.__tAxis.get(),
t_square.get())
self.__temp = interp(cp.asarray(standard_full_time),
self.__tAxis - t_mean, t_square).get()
self.__temp = self.__temp / cp.sum(self.__temp)
self.__tAxis = standard_full_time
self.__is_low_res = True
self.__streakedspectra = np.asarray(
(xuvonly.get(), streaked1.get(), streaked2.get()))
self.__streakspeed = streakspeed_in_meV_per_fs
self.__tAxis = standard_full_time
if discretized:
streaked1 = self.discretized_spectrum(streaked1,
self.num_electrons1)
streaked2 = self.discretized_spectrum(streaked2,
self.num_electrons2)
self.__streakspeed = streakspeed_in_meV_per_fs
return cp.asnumpy(xuvonly), cp.asnumpy(streaked1), cp.asnumpy(
streaked2)
elif discretized:
(xuvonly, streaked1, streaked2) = self.__streakedspectra
streaked1 = self.discretized_spectrum(streaked1,
self.num_electrons1)
streaked2 = self.discretized_spectrum(streaked2,
self.num_electrons2)
return cp.asnumpy(xuvonly), cp.asnumpy(streaked1), cp.asnumpy(
streaked2)
else:
return self.__streakedspectra.copy()
def _index_or_values_interpolation(column, index=None):
"""
Interpolate over a float column. assumes a linear interpolation
strategy using the index of the data to denote spacing of the x
values. For example the data and index [1.0, NaN, 4.0], [1, 3, 4]
would result in [1.0, 3.0, 4.0]
"""
# figure out where the nans are
mask = cp.isnan(column)
# trivial cases, all nan or no nans
num_nan = mask.sum()
if num_nan == 0 or num_nan == len(column):
return column
to_interp = Frame(data={None: column}, index=index)
known_x_and_y = to_interp._apply_boolean_mask(as_column(~mask))
known_x = known_x_and_y._index._column.values
known_y = known_x_and_y._data.columns[0].values
result = cp.interp(to_interp._index.values, known_x, known_y)
# find the first nan
first_nan_idx = (mask == 0).argmax().item()
result[:first_nan_idx] = np.nan
return result
def resampled_TOF_signal(self, TOF_signal):
from numpy import take_along_axis, interp
sorted_TOF_signal = take_along_axis(TOF_signal,
self.TOF_times_sort_order,
axis=0)
resampled_TOF_signal = interp(Raw_data.TOF_times,
self.sorted_TOF_times, sorted_TOF_signal)
return resampled_TOF_signal
def resampled_VLS_signal(self, VLS_signal):
from numpy import take_along_axis, interp
sorted_VLS_signal = take_along_axis(VLS_signal,
self.VLS_pixels_sort_order,
axis=0)
resampled_VLS_signal = interp(Raw_data.VLS_pixels,
self.sorted_VLS_pixels,
sorted_VLS_signal)
return resampled_VLS_signal
def _match_cumulative_cdf(source, template):
"""
Return modified source array so that the cumulative density function of
its values matches the cumulative density function of the template.
"""
src_values, src_unique_indices, src_counts = cp.unique(source.ravel(),
return_inverse=True,
return_counts=True)
tmpl_values, tmpl_counts = cp.unique(template.ravel(), return_counts=True)
# calculate normalized quantiles for each array
src_quantiles = cp.cumsum(src_counts) / source.size
tmpl_quantiles = cp.cumsum(tmpl_counts) / template.size
interp_a_values = cp.interp(src_quantiles, tmpl_quantiles, tmpl_values)
return interp_a_values[src_unique_indices].reshape(source.shape)
def GetSASE_gpu(CentralEnergy, dE_FWHM, dt_FWHM, samples=0, onlyT=False):
from cupy import interp
import cupy as cp
h = 4.135667662 #in eV*fs
dE = dE_FWHM / 2.355 #in eV, converts to sigma
dt = dt_FWHM / 2.355 #in fs, converts to sigma
if samples == 0:
samples = int(600. * dt * CentralEnergy / h)
else:
if (samples < 400. * dt * CentralEnergy / h):
print(
"Number of samples is a little small, proceeding anyway. Got",
samples, "prefer more than", 400. * dt * CentralEnergy / h)
EnAxis = cp.linspace(0.,
20. * CentralEnergy,
num=samples,
dtype=cp.float32)
newEaxis = cp.linspace(0, 140, 32 * 1024)
# EnInput=cp.zeros(samples, dtype=cp.complex64)
# for i in range(samples):
EnInput = cp.exp(-(EnAxis - CentralEnergy)**2 / 2. / dE**2 +
2 * cp.pi * 1j * cp.random.random(size=samples),
dtype=cp.complex64)
EnInput = interp(newEaxis, EnAxis, EnInput)
newTaxis = cp.fft.fftfreq(32 * 1024, d=140 / (32 * 1024)) * h
TOutput = cp.exp(-newTaxis**2 / 2. / dt**2) * cp.fft.fft(EnInput)
# sort TOutput and newTaxis
ind = cp.argsort(newTaxis, axis=0)
newTaxis = cp.sort(newTaxis)
TOutput = cp.take_along_axis(TOutput, ind, axis=0)
# En_FFT=cp.fft.fft(EnInput)
# TAxis=cp.fft.fftfreq(samples,d=(20.*CentralEnergy)/samples)*h
# TOutput=cp.exp(-TAxis**2/2./dt**2)*En_FFT
if not (onlyT):
EnOutput = cp.fft.ifft(TOutput)
if (onlyT):
return newTaxis, TOutput
else:
return newEaxis, EnOutput, newTaxis, TOutput
def equalize_hist(image, nbins=256, mask=None):
"""Return image after histogram equalization.
Parameters
----------
image : array
Image array.
nbins : int, optional
Number of bins for image histogram. Note: this argument is
ignored for integer images, for which each integer is its own
bin.
mask: ndarray of bools or 0s and 1s, optional
Array of same shape as `image`. Only points at which mask == True
are used for the equalization, which is applied to the whole image.
Returns
-------
out : float array
Image array after histogram equalization.
Notes
-----
This function is adapted from [1]_ with the author's permission.
References
----------
.. [1] http://www.janeriksolem.net/histogram-equalization-with-python-and.html
.. [2] https://en.wikipedia.org/wiki/Histogram_equalization
"""
if mask is not None:
mask = cp.array(mask, dtype=bool)
cdf, bin_centers = cumulative_distribution(image[mask], nbins)
else:
cdf, bin_centers = cumulative_distribution(image, nbins)
if False:
out = cp.interp(image.flat, bin_centers, cdf)
else:
# TODO: grlee77: no cp.interp, so have to transfer
out = cp.asarray(
np.interp(image.get().flat, bin_centers.get(), cdf.get())
)
return out.reshape(image.shape)
def _match_cumulative_cdf(source, template):
"""
Return modified source array so that the cumulative density function of
its values matches the cumulative density function of the template.
"""
src_values, src_unique_indices, src_counts = cp.unique(source.ravel(),
return_inverse=True,
return_counts=True)
tmpl_values, tmpl_counts = cp.unique(template.ravel(), return_counts=True)
# calculate normalized quantiles for each array
src_quantiles = cp.cumsum(src_counts) / source.size
tmpl_quantiles = cp.cumsum(tmpl_counts) / template.size
# TODO: grlee77: cp.interp does not exist, so have to transfer to CPU
if not hasattr(cp, "interp"):
src_quantiles = src_quantiles.get()
tmpl_quantiles = tmpl_quantiles.get()
tmpl_values = tmpl_values.get()
interp_a_values = cp.asarray(
np.interp(src_quantiles, tmpl_quantiles, tmpl_values))
else:
interp_a_values = cp.interp(src_quantiles, tmpl_quantiles, tmpl_values)
return interp_a_values[src_unique_indices].reshape(source.shape)
def _calibrate(self, img, findCarrier = True, useCupy = False):
assert len(img) > 2
self.N = len(img[0, :, :])
if self.N != self._lastN:
self._allocate_arrays()
''' define k grids '''
self._dx = self.pixelsize / self.magnification # Sampling in image plane
self._res = self.wavelength / (2 * self.NA)
self._oversampling = self._res / self._dx
self._dk = self._oversampling / (self.N / 2) # Sampling in frequency plane
self._k = np.arange(-self._dk * self.N / 2, self._dk * self.N / 2, self._dk, dtype=np.double)
self._dx2 = self._dx / 2
self._kr = np.sqrt(self._k ** 2 + self._k[:,np.newaxis] ** 2, dtype=np.single)
kxbig = np.arange(-self._dk * self.N, self._dk * self.N, self._dk, dtype=np.single)
kybig = kxbig[:,np.newaxis]
'''Sum input images if there are more than 3'''
if len(img) > 3:
imgs = np.zeros((3, self.N, self.N), dtype=np.single)
for i in range(3):
imgs[i, :, :] = np.sum(img[i:(len(img) // 3) * 3:3, :, :], 0, dtype = np.single)
else:
imgs = np.single(img)
'''Separate bands into DC and 1 high frequency band'''
M = exp(1j * 2 * pi / 3) ** ((np.arange(0, 2)[:, np.newaxis]) * np.arange(0, 3))
sum_prepared_comp = np.zeros((2, self.N, self.N), dtype=np.complex64)
wienerfilter = np.zeros((2 * self.N, 2 * self.N), dtype=np.single)
for k in range(0, 2):
for l in range(0, 3):
sum_prepared_comp[k, :, :] = sum_prepared_comp[k, :, :] + imgs[l, :, :] * M[k, l]
if findCarrier:
# minimum search radius in k-space
mask1 = (self._kr > 1.9 * self.eta)
if useCupy:
self.kx, self.ky = self._coarseFindCarrier_cupy(sum_prepared_comp[0, :, :],
sum_prepared_comp[1, :, :], mask1)
else:
self.kx, self.ky = self._coarseFindCarrier(sum_prepared_comp[0, :, :],
sum_prepared_comp[1, :, :], mask1)
if useCupy:
ckx, cky, p, ampl = self._refineCarrier_cupy(sum_prepared_comp[0, :, :],
sum_prepared_comp[1, :, :], self.kx, self.ky)
else:
ckx, cky, p, ampl = self._refineCarrier(sum_prepared_comp[0, :, :],
sum_prepared_comp[1, :, :], self.kx, self.ky)
self.kx = ckx # store found kx, ky, p and ampl values
self.ky = cky
self.p = p
self.ampl = ampl
if self.debug:
print(f'kx = {ckx}')
print(f'ky = {cky}')
print(f'p = {p}')
print(f'a = {ampl}')
ph = np.single(2 * pi * self.NA / self.wavelength)
xx = np.arange(-self._dx2 * self.N, self._dx2 * self.N, self._dx2, dtype=np.single)
yy = xx
if self.usemodulation:
A = ampl
else:
A = 1
for idx_p in range(0, 3):
pstep = idx_p * 2 * pi / 3
if useCupy:
self._reconfactor[idx_p, :, :] = (
1 + 4 / A * cp.outer(cp.exp(cp.asarray(1j * (ph * cky * yy - pstep + p))),
cp.exp(cp.asarray(1j * ph * ckx * xx))).real).get()
else:
self._reconfactor[idx_p, :, :] = (
1 + 4 / A * np.outer(np.exp(1j * (ph * cky * yy - pstep + p)),
np.exp(1j * ph * ckx * xx)).real)
# calculate pre-filter factors
mask2 = (self._kr < 2)
self._prefilter = np.single((self._tfm(self._kr, mask2) * self._attm(self._kr, mask2)))
self._prefilter = fft.fftshift(self._prefilter)
mtot = np.full((2 * self.N, 2 * self.N), False)
krbig = sqrt((kxbig - ckx) ** 2 + (kybig - cky) ** 2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = wienerfilter[mask] + (self._tf(krbig[mask]) ** 2) * self._att(krbig[mask])
krbig = sqrt((kxbig + ckx) ** 2 + (kybig + cky) ** 2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = wienerfilter[mask] + (self._tf(krbig[mask]) ** 2) * self._att(krbig[mask])
krbig = sqrt(kxbig ** 2 + kybig ** 2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = wienerfilter[mask] + (self._tf(krbig[mask]) ** 2) * self._att(krbig[mask])
self.wienerfilter = wienerfilter
if self.debug:
plt.figure()
plt.title('WienerFilter')
plt.imshow(wienerfilter)
th = np.linspace(0, 2 * pi, 360, dtype = np.single)
inv = np.geterr()['invalid']
np.seterr(invalid = 'ignore')
kmaxth = np.fmax(2, np.fmax(ckx * np.cos(th) + cky * np.sin(th) +
np.sqrt(4 - (ckx * np.sin(th)) ** 2 - (cky * np.cos(th)) ** 2 +
ckx * cky * np.sin(2 * th)),
- ckx * np.cos(th) - cky * np.sin(th) +
np.sqrt(4 - (ckx * np.sin(th)) ** 2 - (cky * np.cos(th)) ** 2 +
ckx * cky * np.sin(2 * th))))
np.seterr(invalid = inv)
if useCupy and 'interp' in dir(cp): # interp not available in cupy version < 9.0.0
kmax = cp.interp(cp.arctan2(cp.asarray(kybig), cp.asarray(kxbig)), cp.asarray(th), cp.asarray(kmaxth), period=2 * pi).astype(np.single).get()
else:
kmax = np.interp(np.arctan2(kybig, kxbig), th, kmaxth, period=2 * pi).astype(np.single)
wienerfilter = mtot * (1 - krbig * mtot / kmax) / (wienerfilter * mtot + self.w ** 2)
self._postfilter = fft.fftshift(wienerfilter)
if opencv:
self._reconfactorU = [cv2.UMat(self._reconfactor[idx_p, :, :]) for idx_p in range(0, 3)]
self._prefilter_ocv = np.single(cv2.dft(fft.ifft2(self._prefilter).real))
pf = np.zeros((self.N, self.N, 2), dtype=np.single)
pf[:, :, 0] = self._prefilter
pf[:, :, 1] = self._prefilter
self._prefilter_ocvU = cv2.UMat(np.single(pf))
self._postfilter_ocv = np.single(cv2.dft(fft.ifft2(self._postfilter).real))
pf = np.zeros((2 * self.N, 2 * self.N, 2), dtype=np.single)
pf[:, :, 0] = self._postfilter
pf[:, :, 1] = self._postfilter
self._postfilter_ocvU = cv2.UMat(np.single(pf))
if cupy:
self._postfilter_cp = cp.asarray(self._postfilter)
def _calibrate(self, img, findCarrier=True, useCupy=False):
assert len(img) > 6
self.N = len(img[0, :, :])
if self.N != self._lastN:
self._allocate_arrays()
''' define k grids '''
self._dx = self.pixelsize / self.magnification # Sampling in image plane
self._res = self.wavelength / (2 * self.NA)
self._oversampling = self._res / self._dx
self._dk = self._oversampling / (self.N / 2
) # Sampling in frequency plane
self._k = np.arange(-self._dk * self.N / 2,
self._dk * self.N / 2,
self._dk,
dtype=np.double)
self._dx2 = self._dx / 2
self._kr = np.sqrt(self._k**2 + self._k[:, np.newaxis]**2,
dtype=np.single)
kxbig = np.arange(-self._dk * self.N,
self._dk * self.N,
self._dk,
dtype=np.single)
kybig = kxbig[:, np.newaxis]
'''Sum input images if there are more than 7'''
if len(img) > 7:
imgs = np.zeros((7, self.N, self.N), dtype=np.single)
for i in range(7):
imgs[i, :, :] = np.sum(img[i:(len(img) // 7) * 7:7, :, :],
0,
dtype=np.single)
else:
imgs = np.single(img)
'''Separate bands into DC and 3 high frequency bands'''
M = np.complex64(
exp(1j * 2 * pi / 7)**((np.arange(0, 4)[:, np.newaxis]) *
np.arange(0, 7)))
wienerfilter = np.zeros((2 * self.N, 2 * self.N), dtype=np.single)
if useCupy:
sum_prepared_comp = cp.dot(cp.asarray(M),
cp.asarray(imgs).transpose(
(1, 0, 2))).get()
else:
sum_prepared_comp = np.zeros((4, self.N, self.N),
dtype=np.complex64)
for k in range(0, 4):
for l in range(0, 7):
sum_prepared_comp[k, :, :] = sum_prepared_comp[
k, :, :] + imgs[l, :, :] * M[k, l]
# find parameters
ckx = np.zeros((3, 1), dtype=np.single)
cky = np.zeros((3, 1), dtype=np.single)
p = np.zeros((3, 1), dtype=np.single)
ampl = np.zeros((3, 1), dtype=np.single)
if findCarrier:
# minimum search radius in k-space
mask1 = (self._kr > 1.9 * self.eta)
for i in range(0, 3):
if useCupy:
self.kx[i], self.ky[i] = self._coarseFindCarrier_cupy(
sum_prepared_comp[0, :, :],
sum_prepared_comp[i + 1, :, :], mask1)
else:
self.kx[i], self.ky[i] = self._coarseFindCarrier(
sum_prepared_comp[0, :, :],
sum_prepared_comp[i + 1, :, :], mask1)
for i in range(0, 3):
if useCupy:
ckx[i], cky[i], p[i], ampl[i] = self._refineCarrier_cupy(
sum_prepared_comp[0, :, :], sum_prepared_comp[i + 1, :, :],
self.kx[i], self.ky[i])
else:
ckx[i], cky[i], p[i], ampl[i] = self._refineCarrier(
sum_prepared_comp[0, :, :], sum_prepared_comp[i + 1, :, :],
self.kx[i], self.ky[i])
self.kx = ckx # store found kx, ky, p and ampl values
self.ky = cky
self.p = p
self.ampl = ampl
if self.debug:
print(f'kx = {ckx[0]}, {ckx[1]}, {ckx[2]}')
print(f'ky = {cky[0]}, {cky[1]}, {cky[2]}')
print(f'p = {p[0]}, {p[1]}, {p[2]}')
print(f'a = {ampl[0]}, {ampl[1]}, {ampl[2]}')
ph = np.single(2 * pi * self.NA / self.wavelength)
xx = np.arange(-self._dx2 * self.N,
self._dx2 * self.N,
self._dx2,
dtype=np.single)
yy = xx
if self.usemodulation:
A = [float(ampl[i]) for i in range(3)]
else:
if self.axial:
A = [6.0 for i in range(3)]
else:
A = [12.0 for i in range(3)]
for idx_p in range(0, 7):
pstep = idx_p * 2 * pi / 7
if useCupy:
self._reconfactor[idx_p, :, :] = (
1 + 4 / A[0] * cp.outer(
cp.exp(
cp.asarray(1j *
(ph * cky[0] * yy - pstep + p[0]))),
cp.exp(cp.asarray(1j * ph * ckx[0] * xx))).real +
4 / A[1] * cp.outer(
cp.exp(
cp.asarray(1j *
(ph * cky[1] * yy - 2 * pstep + p[1]))),
cp.exp(cp.asarray(1j * ph * ckx[1] * xx))).real +
4 / A[2] * cp.outer(
cp.exp(
cp.asarray(1j *
(ph * cky[2] * yy - 3 * pstep + p[2]))),
cp.exp(cp.asarray(1j * ph * ckx[2] * xx))).real).get()
else:
self._reconfactor[idx_p, :, :] = (1 + 4 / A[0] * np.outer(
np.exp(1j * (ph * cky[0] * yy - pstep + p[0])),
np.exp(1j * ph * ckx[0] * xx)).real + 4 / A[1] * np.outer(
np.exp(1j * (ph * cky[1] * yy - 2 * pstep + p[1])),
np.exp(
1j * ph * ckx[1] * xx)).real + 4 / A[2] * np.outer(
np.exp(1j *
(ph * cky[2] * yy - 3 * pstep + p[2])),
np.exp(1j * ph * ckx[2] * xx)).real)
# calculate pre-filter factors
mask2 = (self._kr < 2)
self._prefilter = np.single(
(self._tfm(self._kr, mask2) * self._attm(self._kr, mask2)))
self._prefilter = fft.fftshift(self._prefilter)
mtot = np.full((2 * self.N, 2 * self.N), False)
th = np.linspace(0, 2 * pi, 360, dtype=np.single)
inv = np.geterr()['invalid']
kmaxth = 2
for i in range(0, 3):
krbig = sqrt((kxbig - ckx[i])**2 + (kybig - cky[i])**2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = wienerfilter[mask] + (self._tf(
krbig[mask])**2) * self._att(krbig[mask])
krbig = sqrt((kxbig + ckx[i])**2 + (kybig + cky[i])**2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = wienerfilter[mask] + (self._tf(
krbig[mask])**2) * self._att(krbig[mask])
np.seterr(invalid='ignore'
) # Silence sqrt warnings for kmaxth calculations
kmaxth = np.fmax(
kmaxth,
np.fmax(
ckx[i] * np.cos(th) + cky[i] * np.sin(th) +
np.sqrt(4 - (ckx[i] * np.sin(th))**2 -
(cky[i] * np.cos(th))**2 +
ckx[i] * cky[i] * np.sin(2 * th)),
-ckx[i] * np.cos(th) - cky[i] * np.sin(th) +
np.sqrt(4 - (ckx[i] * np.sin(th))**2 -
(cky[i] * np.cos(th))**2 +
ckx[i] * cky[i] * np.sin(2 * th))))
np.seterr(invalid=inv)
if self.debug:
plt.figure()
plt.plot(th, kmaxth)
krbig = sqrt(kxbig**2 + kybig**2)
mask = (krbig < 2)
mtot = mtot | mask
wienerfilter[mask] = (
wienerfilter[mask] +
self._tf(krbig[mask])**2 * self._att(krbig[mask]))
self.wienerfilter = wienerfilter
if useCupy and 'interp' in dir(
cp): # interp not available in cupy version < 9.0.0
kmax = cp.interp(cp.arctan2(cp.asarray(kybig), cp.asarray(kxbig)),
cp.asarray(th),
cp.asarray(kmaxth),
period=2 * pi).astype(np.single).get()
else:
kmax = np.interp(np.arctan2(kybig, kxbig),
th,
kmaxth,
period=2 * pi).astype(np.single)
if self.debug:
plt.figure()
plt.title('WienerFilter')
plt.imshow(wienerfilter)
wienerfilter = mtot * (1 - krbig * mtot / kmax) / (
wienerfilter * mtot + self.w**2)
self._postfilter = fft.fftshift(wienerfilter)
if self.cleanup:
imgo = self.reconstruct_fftw(img)
kernel = np.ones((5, 5), np.uint8)
mask_tmp = abs(fft.fftshift(fft.fft2(imgo))) > (
10 * gaussian_filter(abs(fft.fftshift(fft.fft2(imgo))), 5))
mask = scipy.ndimage.morphology.binary_dilation(
np.single(mask_tmp), kernel)
mask[self.N - 12:self.N + 13, self.N - 12:self.N + 13] = np.full(
(25, 25), False)
mask_shift = (fft.fftshift(mask))
self._postfilter[mask_shift.astype(bool)] = 0
if opencv:
self._reconfactorU = [
cv2.UMat(self._reconfactor[idx_p, :, :])
for idx_p in range(0, 7)
]
self._prefilter_ocv = np.single(
cv2.dft(fft.ifft2(self._prefilter).real))
pf = np.zeros((self.N, self.N, 2), dtype=np.single)
pf[:, :, 0] = self._prefilter
pf[:, :, 1] = self._prefilter
self._prefilter_ocvU = cv2.UMat(np.single(pf))
self._postfilter_ocv = np.single(
cv2.dft(fft.ifft2(self._postfilter).real))
pf = np.zeros((2 * self.N, 2 * self.N, 2), dtype=np.single)
pf[:, :, 0] = self._postfilter
pf[:, :, 1] = self._postfilter
self._postfilter_ocvU = cv2.UMat(np.single(pf))
if cupy:
self._postfilter_cp = cp.asarray(self._postfilter)
def __call__(self, x):
return xp.interp(xp.asarray(x), self.xx, self.yy)
