def make_projection(shape,
ps,
axis,
mesh,
center,
lamino_angle,
tomo_angle,
ss=1):
from syris.imageprocessing import bin_image
if axis == 'z':
lamino_angle = lamino_angle + 90 * q.deg
tomo_angle = -tomo_angle
axis = Y_AX if axis == 'y' else Z_AX
mesh.clear_transformation()
mesh.translate(center)
mesh.rotate(lamino_angle, X_AX)
mesh.rotate(tomo_angle, axis)
orig_shape = shape
shape = tuple([n * ss for n in orig_shape])
ps = ps / ss
# t=None for trajectory override
projection = mesh.project(shape, ps, t=None)
if ss > 1:
projection = bin_image(projection, orig_shape, average=True)
return projection.get()
def get_image(self,
photons,
shot_noise=True,
amplifier_noise=True,
psf=True,
queue=None):
"""Get digital counts image from incoming *photons*. The resulting image is based on the
incoming photons and dark current. We apply noise based on EMVA 1288 standard according to
which the variance :math:`\sigma_y^2 = K^2 ( \sigma_e^2 + \sigma_d^2 ) + \sigma_q^2`, where
:math:`K` is the system gain, :math:`\sigma_e^2` is the poisson- distributed shot noise
variance, :math:`\sigma_d^2` is the normal distributed electronics noise variance and
:math:`\sigma_q^2` is the quantization noise variance. If *shot_noise* is False don't apply
it. If *amplifier_noise* is False don't apply it as well. If *psf* is False don't apply the
point spread function.
"""
if self._last_input_shape != photons.shape:
self._last_input_shape = photons.shape
self._bin_factor = (photons.shape[0] / self.shape[0],
photons.shape[1] / self.shape[1])
if queue is None:
queue = cfg.OPENCL.queue
# Shot noise
# Adjust dark current for later binning and gain
dark = float(
self.dark_current) / self._bin_factor[0] / self._bin_factor[1]
electrons = dark + gutil.get_host(photons)
if self._bin_factor != (1, 1):
if psf:
sigma = (fwnm_to_sigma(self._bin_factor[0]),
fwnm_to_sigma(self._bin_factor[1]))
small = decimate(electrons,
self.shape,
sigma=sigma,
queue=queue)
else:
small = bin_image(electrons, self.shape, queue=queue)
electrons = gutil.get_host(small)
if shot_noise:
electrons = np.random.poisson(electrons)
if amplifier_noise and self.amplifier_sigma > 0:
# Add electronics noise
electrons = np.random.normal(electrons, self.amplifier_sigma)
counts = self.gain * electrons
# Cut the values beyond the maximum represented grey value given by
# bytes per pixel.
counts[counts > self.max_grey_value] = self.max_grey_value
# Apply quantization noise
return counts.astype(self.dtype)
def test_sum(self):
m = 8
n = 16
image = np.arange(m * n).reshape(m, n).astype(cfg.PRECISION.np_float)
cl_im = cl_array.to_device(cfg.OPENCL.queue, image)
sizes = (1, 2, 4)
for shape in itertools.product(sizes, sizes):
region = (m // shape[0], n // shape[1])
gt = bin_cpu(image, shape)
res = ip.bin_image(cl_im, shape)
np.testing.assert_equal(gt, res)
# Test averaging
res = ip.bin_image(cl_im, shape, average=True)
np.testing.assert_equal(gt / (region[0] * region[1]), res)
# Not a divisor
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (4, 10))
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (5, 8))
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (4, 7), offset=(2, 2))
def test_sum(self):
m = 8
n = 16
image = np.arange(m * n).reshape(m, n).astype(cfg.PRECISION.np_float)
cl_im = cl_array.to_device(cfg.OPENCL.queue, image)
sizes = (1, 2, 4)
for shape in itertools.product(sizes, sizes):
region = (m / shape[0], n / shape[1])
gt = bin_cpu(image, shape)
res = ip.bin_image(cl_im, shape)
np.testing.assert_equal(gt, res)
# Test averaging
res = ip.bin_image(cl_im, shape, average=True)
np.testing.assert_equal(gt / (region[0] * region[1]), res)
# Not a divisor
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (4, 10))
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (5, 8))
self.assertRaises(RuntimeError, ip.bin_image, cl_im, (4, 7), offset=(2, 2))
def make_ground_truth(args, shape, mesh):
"""Shape is (y, x), so the total number of slices is y."""
import syris.config as cfg
from syris.imageprocessing import bin_image
if args.z_chunk % args.supersampling:
raise ValueError('z_chunk must be dividable by supersampling')
queue = cfg.OPENCL.queue
# Move the mesh to the middle
ps = args.pixel_size / args.supersampling
psm = ps.simplified.magnitude
orig_shape = shape
shape = tuple([n * args.supersampling for n in shape])
# Make sure the projections are computed with the same x- and y-offsets
point = (shape[1] * psm / 2, shape[0] * psm / 2, shape[1] * psm / 2) * q.m
LOG.info('Mesh shift: {}'.format(point.rescale(q.um)))
LOG.info('Mesh shift in pixels: {}'.format(
(point / args.pixel_size).simplified.magnitude))
mesh.translate(point)
mesh.transform()
mesh.sort()
z_stack = np.empty((args.supersampling, ) + orig_shape,
dtype=cfg.PRECISION.np_float)
for i in range(0, shape[0], args.z_chunk):
end = min(i + args.z_chunk, shape[0])
offset = (0, i * ps.rescale(q.um).magnitude, 0) * q.um
slices = mesh.compute_slices((end - i, ) + shape, ps,
offset=offset).get()
LOG.info('Computing slices {}-{}'.format(i, end))
enumerated = list(enumerate(slices))[::args.supersampling]
for j, sl in enumerated:
# Z-dimension downsampling
for k in range(args.supersampling):
z_stack[k] = bin_image(slices[j + k],
orig_shape,
average=True,
queue=queue).get()
# Sum only the slices which are present (last run might not go to the end)
sl = np.mean(z_stack[:slices.shape[0]], axis=0)
index = (i + j) / args.supersampling
save_image(args.prefix.format(index), sl)
return sl
def get_image(self, photons, shot_noise=True, amplifier_noise=True, psf=True, queue=None):
"""Get digital counts image from incoming *photons*. The resulting image is based on the
incoming photons and dark current. We apply noise based on EMVA 1288 standard according to
which the variance :math:`\sigma_y^2 = K^2 ( \sigma_e^2 + \sigma_d^2 ) + \sigma_q^2`, where
:math:`K` is the system gain, :math:`\sigma_e^2` is the poisson- distributed shot noise
variance, :math:`\sigma_d^2` is the normal distributed electronics noise variance and
:math:`\sigma_q^2` is the quantization noise variance. If *shot_noise* is False don't apply
it. If *amplifier_noise* is False don't apply it as well. If *psf* is False don't apply the
point spread function.
"""
if self._last_input_shape != photons.shape:
self._last_input_shape = photons.shape
self._bin_factor = (photons.shape[0] / self.shape[0], photons.shape[1] / self.shape[1])
if queue is None:
queue = cfg.OPENCL.queue
# Shot noise
# Adjust dark current for later binning and gain
dark = float(self.dark_current) / self._bin_factor[0] / self._bin_factor[1]
electrons = dark + gutil.get_host(photons)
if self._bin_factor != (1, 1):
if psf:
sigma = (fwnm_to_sigma(self._bin_factor[0]), fwnm_to_sigma(self._bin_factor[1]))
small = decimate(electrons, self.shape, sigma=sigma, queue=queue)
else:
small = bin_image(electrons, self.shape, queue=queue)
electrons = gutil.get_host(small)
if shot_noise:
electrons = np.random.poisson(electrons)
if amplifier_noise and self.amplifier_sigma > 0:
# Add electronics noise
electrons = np.random.normal(electrons, self.amplifier_sigma)
counts = self.gain * electrons
# Cut the values beyond the maximum represented grey value given by
# bytes per pixel.
counts[counts > self.max_grey_value] = self.max_grey_value
# Apply quantization noise
return counts.astype(self.dtype)
def make_projection(shape, ps, axis, mesh, center, lamino_angle, tomo_angle, ss=1):
from syris.imageprocessing import bin_image
if axis == 'z':
lamino_angle = lamino_angle + 90 * q.deg
tomo_angle = -tomo_angle
axis = Y_AX if axis == 'y' else Z_AX
mesh.clear_transformation()
mesh.translate(center)
mesh.rotate(lamino_angle, X_AX)
mesh.rotate(tomo_angle, axis)
orig_shape = shape
shape = tuple([n * ss for n in orig_shape])
ps = ps / ss
# t=None for trajectory override
projection = mesh.project(shape, ps, t=None)
if ss > 1:
projection = bin_image(projection, orig_shape, average=True)
return projection.get()
def make_ground_truth(args, shape, mesh):
"""Shape is (y, x), so the total number of slices is y."""
import syris.config as cfg
from syris.imageprocessing import bin_image
if args.z_chunk % args.supersampling:
raise ValueError('z_chunk must be dividable by supersampling')
queue = cfg.OPENCL.queue
# Move the mesh to the middle
ps = args.pixel_size / args.supersampling
psm = ps.simplified.magnitude
orig_shape = shape
shape = tuple([n * args.supersampling for n in shape])
# Make sure the projections are computed with the same x- and y-offsets
point = (shape[1] * psm / 2, shape[0] * psm / 2, shape[1] * psm / 2) * q.m
LOG.info('Mesh shift: {}'.format(point.rescale(q.um)))
LOG.info('Mesh shift in pixels: {}'.format((point / args.pixel_size).simplified.magnitude))
mesh.translate(point)
mesh.transform()
mesh.sort()
z_stack = np.empty((args.supersampling,) + orig_shape, dtype=cfg.PRECISION.np_float)
for i in range(0, shape[0], args.z_chunk):
end = min(i + args.z_chunk, shape[0])
offset = (0, i * ps.rescale(q.um).magnitude, 0) * q.um
slices = mesh.compute_slices((end - i,) + shape, ps, offset=offset).get()
LOG.info('Computing slices {}-{}'.format(i, end))
enumerated = list(enumerate(slices))[::args.supersampling]
for j, sl in enumerated:
# Z-dimension downsampling
for k in range(args.supersampling):
z_stack[k] = bin_image(slices[j + k], orig_shape, average=True, queue=queue).get()
# Sum only the slices which are present (last run might not go to the end)
sl = np.mean(z_stack[:slices.shape[0]], axis=0)
index = (i + j) / args.supersampling
save_image(args.prefix.format(index), sl)
return sl
