__author__ = 'Clemens Prescher'
from pyFAI.blob_detection import BlobDetection
from Data.ImgData import ImgData
import numpy as np
import pylab
img_data = ImgData()
# img_data.load('/Users/Doomgoroth/Programming/Large Projects/Dioptas/Testing/pyFAITest/17_LaB6_dc300-00000.tif')
img_data.load(
'/Users/Doomgoroth/Programming/Large Projects/Dioptas/Testing/pyFAITest/LaB6_WOS_30keV_005.tif'
)
bd = BlobDetection(np.log1p(img_data.get_img_data()))
bd.process()
x = []
y = []
int = []
sigma = []
print bd.keypoints.__len__()
for j in range(bd.keypoints.__len__()):
k = bd.keypoints[j]
int.append(k[2])
sigma.append(k[3])
if sigma[-1] > 0.25:
x.append(k[0])
y.append(k[1])
pylab.hist(int)
class CalibrationData(object):
def __init__(self, img_data=None):
self.img_data = img_data
self.points = []
self.points_index = []
self.geometry = AzimuthalIntegrator()
self.calibrant = Calibrant()
self.start_values = {'dist': 200e-3,
'wavelength': 0.3344e-10,
'pixel_width': 79e-6,
'pixel_height': 79e-6,
'polarization_factor': 0.95}
self.fit_wavelength = False
self.is_calibrated = False
self.use_mask = False
self.calibration_name = 'None'
self.polarization_factor = 0.95
self._calibrants_working_dir = os.path.dirname(Calibrants.__file__)
def find_peaks_automatic(self, x, y, peak_ind):
massif = Massif(self.img_data.img_data)
cur_peak_points = massif.find_peaks([x, y])
if len(cur_peak_points):
self.points.append(np.array(cur_peak_points))
self.points_index.append(peak_ind)
return np.array(cur_peak_points)
def find_peak(self, x, y, search_size, peak_ind):
left_ind = np.round(x - search_size * 0.5)
top_ind = np.round(y - search_size * 0.5)
x_ind, y_ind = np.where(self.img_data.img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)] == \
self.img_data.img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)].max())
x_ind = x_ind[0] + left_ind
y_ind = y_ind[0] + top_ind
self.points.append(np.array([x_ind, y_ind]))
self.points_index.append(peak_ind)
return np.array([np.array((x_ind, y_ind))])
def clear_peaks(self):
self.points = []
self.points_index = []
def setup_peak_search_algorithm(self, algorithm, mask=None):
# init the peak search algorithm
if algorithm == 'Massif':
self.peak_search_algorithm = Massif(self.img_data.img_data)
elif algorithm == 'Blob':
if mask is not None:
self.peak_search_algorithm = BlobDetection(self.img_data.img_data * mask)
else:
self.peak_search_algorithm = BlobDetection(self.img_data.img_data)
self.peak_search_algorithm.process()
else:
return
def search_peaks_on_ring(self, peak_index, delta_tth=0.1, min_mean_factor=1,
upper_limit=55000, mask=None):
if not self.is_calibrated:
return
#transform delta from degree into radians
delta_tth = delta_tth / 180.0 * np.pi
# get appropiate two theta value for the ring number
tth_calibrant_list = self.calibrant.get_2th()
tth_calibrant = np.float(tth_calibrant_list[peak_index])
# get the calculated two theta values for the whole image
if self.geometry._ttha is None:
tth_array = self.geometry.twoThetaArray(self.img_data.img_data.shape)
else:
tth_array = self.geometry._ttha
# create mask based on two_theta position
ring_mask = abs(tth_array - tth_calibrant) <= delta_tth
if mask is not None:
mask = np.logical_and(ring_mask, np.logical_not(mask))
else:
mask = ring_mask
# calculate the mean and standard deviation of this area
sub_data = np.array(self.img_data.img_data.ravel()[np.where(mask.ravel())], dtype=np.float64)
sub_data[np.where(sub_data > upper_limit)] = np.NaN
mean = np.nanmean(sub_data)
std = np.nanstd(sub_data)
# set the threshold into the mask (don't detect very low intensity peaks)
threshold = min_mean_factor * mean + std
mask2 = np.logical_and(self.img_data.img_data > threshold, mask)
mask2[np.where(self.img_data.img_data > upper_limit)] = False
size2 = mask2.sum(dtype=int)
keep = int(np.ceil(np.sqrt(size2)))
try:
res = self.peak_search_algorithm.peaks_from_area(mask2, Imin=mean - std, keep=keep)
except IndexError:
res = []
# Store the result
if len(res):
self.points.append(np.array(res))
self.points_index.append(peak_index)
def set_calibrant(self, filename):
self.calibrant = Calibrant()
self.calibrant.load_file(filename)
self.geometry.calibrant = self.calibrant
def set_start_values(self, start_values):
self.start_values = start_values
self.polarization_factor = start_values['polarization_factor']
def calibrate(self):
self.geometry = GeometryRefinement(self.create_point_array(self.points, self.points_index),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'],
calibrant=self.calibrant)
self.refine()
self.integrate()
self.is_calibrated = True
self.calibration_name = 'current'
def refine(self):
self.geometry.data = self.create_point_array(self.points, self.points_index)
self.geometry.refine2()
if self.fit_wavelength:
self.geometry.refine2_wavelength(fix=[])
def integrate(self):
self.integrate_1d()
self.integrate_2d()
def integrate_1d(self, num_points=1400, mask=None, polarization_factor=None, filename=None, unit='2th_deg'):
if np.sum(mask) == self.img_data.img_data.shape[0] * self.img_data.img_data.shape[1]:
#do not perform integration if the image is completelye masked...
return self.tth, self.int
if polarization_factor is None:
polarization_factor = self.polarization_factor
if unit is 'd_A':
self.tth, self.int = self.geometry.integrate1d(self.img_data.img_data, num_points, method='lut',
unit='2th_deg',
mask=mask, polarization_factor=polarization_factor,
filename=filename)
ind = np.where(self.tth > 0)
self.tth = self.geometry.wavelength / (2 * np.sin(self.tth[ind] / 360 * np.pi)) * 1e10
self.int = self.int[ind]
else:
self.tth, self.int = self.geometry.integrate1d(self.img_data.img_data, num_points, method='lut', unit=unit,
mask=mask, polarization_factor=polarization_factor,
filename=filename)
if self.int.max() > 0:
ind = np.where(self.int > 0)
self.tth = self.tth[ind]
self.int = self.int[ind]
return self.tth, self.int
def integrate_2d(self, mask=None, polarization_factor=None, unit='2th_deg'):
if polarization_factor is None:
polarization_factor = self.polarization_factor
res = self.geometry.integrate2d(self.img_data.img_data, 2048, 2048, method='lut', mask=mask, unit=unit,
polarization_factor=polarization_factor)
self.cake_img = res[0]
self.cake_tth = res[1]
self.cake_azi = res[2]
return self.cake_img
def create_point_array(self, points, points_ind):
res = []
for i, point_list in enumerate(points):
if point_list.shape == (2,):
res.append([point_list[0], point_list[1], points_ind[i]])
else:
for point in point_list:
res.append([point[0], point[1], points_ind[i]])
return np.array(res)
def get_point_array(self):
return self.create_point_array(self.points, self.points_index)
def get_calibration_parameter(self):
pyFAI_parameter = self.geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
try:
fit2d_parameter = self.geometry.getFit2D()
fit2d_parameter['polarization_factor'] = self.polarization_factor
except TypeError:
fit2d_parameter = None
try:
pyFAI_parameter['wavelength'] = self.geometry.wavelength
fit2d_parameter['wavelength'] = self.geometry.wavelength
except RuntimeWarning:
pyFAI_parameter['wavelength'] = 0
return pyFAI_parameter, fit2d_parameter
def load(self, filename):
self.geometry = GeometryRefinement(np.zeros((2, 3)),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'])
self.geometry.load(filename)
self.calibration_name = get_base_name(filename)
self.is_calibrated = True
def save(self, filename):
self.geometry.save(filename)
self.calibration_name = get_base_name(filename)
class CalibrationModel(QtCore.QObject):
def __init__(self, img_model=None):
super(CalibrationModel, self).__init__()
"""
:param img_model:
:type img_model: ImgModel
"""
self.img_model = img_model
self.points = []
self.points_index = []
self.pattern_geometry = AzimuthalIntegrator()
self.pattern_geometry_img_shape = None
self.cake_geometry = None
self.cake_geometry_img_shape = None
self.calibrant = Calibrant()
self.pattern_geometry.wavelength = 0.3344e-10
self.start_values = {'dist': 200e-3,
'wavelength': 0.3344e-10,
'pixel_width': 79e-6,
'pixel_height': 79e-6,
'polarization_factor': 0.99}
self.orig_pixel1 = 79e-6
self.orig_pixel2 = 79e-6
self.fit_wavelength = False
self.fit_distance = True
self.fit_poni1 = True
self.fit_poni2 = True
self.fit_rot1 = True
self.fit_rot2 = True
self.fit_rot3 = True
self.is_calibrated = False
self.use_mask = False
self.filename = ''
self.calibration_name = 'None'
self.polarization_factor = 0.99
self.supersampling_factor = 1
self.correct_solid_angle = True
self._calibrants_working_dir = calibrants_path
self.distortion_spline_filename = None
self.tth = np.linspace(0, 25)
self.int = np.sin(self.tth)
self.num_points = len(self.int)
self.cake_img = np.zeros((2048, 2048))
self.cake_tth = None
self.cake_azi = None
self.peak_search_algorithm = None
def find_peaks_automatic(self, x, y, peak_ind):
"""
Searches peaks by using the Massif algorithm
:param float x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param float y:
y-coordinate in pixel - should be from original image (not supersampled y-coordinate)
:param peak_ind:
peak/ring index to which the found points will be added
:return:
array of points found
"""
massif = Massif(self.img_model._img_data)
cur_peak_points = massif.find_peaks((int(np.round(x)), int(np.round(y))), stdout=DummyStdOut())
if len(cur_peak_points):
self.points.append(np.array(cur_peak_points))
self.points_index.append(peak_ind)
return np.array(cur_peak_points)
def find_peak(self, x, y, search_size, peak_ind):
"""
Searches a peak around the x,y position. It just searches for the maximum value in a specific search size.
:param int x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param int y:
y-coordinate in pixel - should be form original image (not supersampled y-coordinate)
:param search_size:
the length of the search rectangle in pixels in all direction in which the algorithm searches for
the maximum peak
:param peak_ind:
peak/ring index to which the found points will be added
:return:
point found (as array)
"""
left_ind = int(np.round(x - search_size * 0.5))
if left_ind < 0:
left_ind = 0
top_ind = int(np.round(y - search_size * 0.5))
if top_ind < 0:
top_ind = 0
search_array = self.img_model.img_data[left_ind:(left_ind + search_size), top_ind:(top_ind + search_size)]
x_ind, y_ind = np.where(search_array == search_array.max())
x_ind = x_ind[0] + left_ind
y_ind = y_ind[0] + top_ind
self.points.append(np.array([x_ind, y_ind]))
self.points_index.append(peak_ind)
return np.array([np.array((x_ind, y_ind))])
def clear_peaks(self):
self.points = []
self.points_index = []
def remove_last_peak(self):
if self.points:
num_points = int(self.points[-1].size/2) # each peak is x, y so length is twice as number of peaks
self.points.pop(-1)
self.points_index.pop(-1)
return num_points
def create_cake_geometry(self):
self.cake_geometry = AzimuthalIntegrator(splineFile=self.distortion_spline_filename)
pyFAI_parameter = self.pattern_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
pyFAI_parameter['wavelength'] = self.pattern_geometry.wavelength
self.cake_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.cake_geometry.wavelength = pyFAI_parameter['wavelength']
def setup_peak_search_algorithm(self, algorithm, mask=None):
"""
Initializes the peak search algorithm on the current image
:param algorithm:
peak search algorithm used. Possible algorithms are 'Massif' and 'Blob'
:param mask:
if a mask is used during the process this is provided here as a 2d array for the image.
"""
if algorithm == 'Massif':
self.peak_search_algorithm = Massif(self.img_model.raw_img_data)
elif algorithm == 'Blob':
if mask is not None:
self.peak_search_algorithm = BlobDetection(self.img_model.raw_img_data * mask)
else:
self.peak_search_algorithm = BlobDetection(self.img_model.raw_img_data)
self.peak_search_algorithm.process()
else:
return
def search_peaks_on_ring(self, ring_index, delta_tth=0.1, min_mean_factor=1,
upper_limit=55000, mask=None):
"""
This function is searching for peaks on an expected ring. It needs an initial calibration
before. Then it will search for the ring within some delta_tth and other parameters to get
peaks from the calibrant.
:param ring_index: the index of the ring for the search
:param delta_tth: search space around the expected position in two theta
:param min_mean_factor: a factor determining the minimum peak intensity to be picked up. it is based
on the mean value of the search area defined by delta_tth. Pick a large value
for larger minimum value and lower for lower minimum value. Therefore, a smaller
number is more prone to picking up noise. typical values like between 1 and 3.
:param upper_limit: maximum intensity for the peaks to be picked
:param mask: in case the image has to be masked from certain areas, it need to be given here. Default is None.
The mask should be given as an 2d array with the same dimensions as the image, where 1 denotes a
masked pixel and all others should be 0.
"""
self.reset_supersampling()
if not self.is_calibrated:
return
# transform delta from degree into radians
delta_tth = delta_tth / 180.0 * np.pi
# get appropriate two theta value for the ring number
tth_calibrant_list = self.calibrant.get_2th()
if ring_index >= len(tth_calibrant_list):
raise NotEnoughSpacingsInCalibrant()
tth_calibrant = np.float(tth_calibrant_list[ring_index])
# get the calculated two theta values for the whole image
tth_array = self.pattern_geometry.twoThetaArray(self.img_model._img_data.shape)
# create mask based on two_theta position
ring_mask = abs(tth_array - tth_calibrant) <= delta_tth
if mask is not None:
mask = np.logical_and(ring_mask, np.logical_not(mask))
else:
mask = ring_mask
# calculate the mean and standard deviation of this area
sub_data = np.array(self.img_model._img_data.ravel()[np.where(mask.ravel())], dtype=np.float64)
sub_data[np.where(sub_data > upper_limit)] = np.NaN
mean = np.nanmean(sub_data)
std = np.nanstd(sub_data)
# set the threshold into the mask (don't detect very low intensity peaks)
threshold = min_mean_factor * mean + std
mask2 = np.logical_and(self.img_model._img_data > threshold, mask)
mask2[np.where(self.img_model._img_data > upper_limit)] = False
size2 = mask2.sum(dtype=int)
keep = int(np.ceil(np.sqrt(size2)))
try:
sys.stdout = DummyStdOut
res = self.peak_search_algorithm.peaks_from_area(mask2, Imin=mean - std, keep=keep)
sys.stdout = sys.__stdout__
except IndexError:
res = []
# Store the result
if len(res):
self.points.append(np.array(res))
self.points_index.append(ring_index)
self.set_supersampling()
self.pattern_geometry.reset()
def set_calibrant(self, filename):
self.calibrant = Calibrant()
self.calibrant.load_file(filename)
self.pattern_geometry.calibrant = self.calibrant
def set_start_values(self, start_values):
self.start_values = start_values
self.polarization_factor = start_values['polarization_factor']
def calibrate(self):
self.pattern_geometry = GeometryRefinement(self.create_point_array(self.points, self.points_index),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'],
calibrant=self.calibrant,
splineFile=self.distortion_spline_filename)
self.orig_pixel1 = self.start_values['pixel_width']
self.orig_pixel2 = self.start_values['pixel_height']
self.refine()
self.create_cake_geometry()
self.is_calibrated = True
self.calibration_name = 'current'
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.pattern_geometry.reset()
def refine(self):
self.reset_supersampling()
self.pattern_geometry.data = self.create_point_array(self.points, self.points_index)
fix = ['wavelength']
if self.fit_wavelength:
fix = []
if not self.fit_distance:
fix.append('dist')
if not self.fit_poni1:
fix.append('poni1')
if not self.fit_poni2:
fix.append('poni2')
if not self.fit_rot1:
fix.append('rot1')
if not self.fit_rot2:
fix.append('rot2')
if not self.fit_rot3:
fix.append('rot3')
if self.fit_wavelength:
self.pattern_geometry.refine2()
self.pattern_geometry.refine2_wavelength(fix=fix)
self.create_cake_geometry()
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.pattern_geometry.reset()
def integrate_1d(self, num_points=None, mask=None, polarization_factor=None, filename=None,
unit='2th_deg', method='csr'):
if np.sum(mask) == self.img_model.img_data.shape[0] * self.img_model.img_data.shape[1]:
# do not perform integration if the image is completely masked...
return self.tth, self.int
if self.pattern_geometry_img_shape != self.img_model.img_data.shape:
# if cake geometry was used on differently shaped image before the azimuthal integrator needs to be reset
self.pattern_geometry.reset()
self.pattern_geometry_img_shape = self.img_model.img_data.shape
if polarization_factor is None:
polarization_factor = self.polarization_factor
if num_points is None:
num_points = self.calculate_number_of_pattern_points(2)
self.num_points = num_points
t1 = time.time()
if unit is 'd_A':
try:
self.tth, self.int = self.pattern_geometry.integrate1d(self.img_model.img_data, num_points,
method=method,
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
correctSolidAngle=self.correct_solid_angle,
filename=filename)
except NameError:
self.tth, self.int = self.pattern_geometry.integrate1d(self.img_model.img_data, num_points,
method='csr',
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
correctSolidAngle=self.correct_solid_angle,
filename=filename)
self.tth = self.pattern_geometry.wavelength / (2 * np.sin(self.tth / 360 * np.pi)) * 1e10
self.int = self.int
else:
try:
self.tth, self.int = self.pattern_geometry.integrate1d(self.img_model.img_data, num_points,
method=method,
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
correctSolidAngle=self.correct_solid_angle,
filename=filename)
except NameError:
self.tth, self.int = self.pattern_geometry.integrate1d(self.img_model.img_data, num_points,
method='csr',
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
correctSolidAngle=self.correct_solid_angle,
filename=filename)
logger.info('1d integration of {0}: {1}s.'.format(os.path.basename(self.img_model.filename), time.time() - t1))
ind = np.where((self.int > 0) & (~np.isnan(self.int)))
self.tth = self.tth[ind]
self.int = self.int[ind]
return self.tth, self.int
def integrate_2d(self, mask=None, polarization_factor=None, unit='2th_deg', method='csr',
rad_points=None, azimuth_points=360,
azimuth_range=None):
if polarization_factor is None:
polarization_factor = self.polarization_factor
if self.cake_geometry_img_shape != self.img_model.img_data.shape:
# if cake geometry was used on differently shaped image before the azimuthal integrator needs to be reset
self.cake_geometry.reset()
self.cake_geometry_img_shape = self.img_model.img_data.shape
if rad_points is None:
rad_points = self.calculate_number_of_pattern_points(2)
self.num_points = rad_points
t1 = time.time()
res = self.cake_geometry.integrate2d(self.img_model.img_data, rad_points, azimuth_points,
azimuth_range=azimuth_range,
method=method,
mask=mask,
unit=unit,
polarization_factor=polarization_factor,
correctSolidAngle=self.correct_solid_angle)
logger.info('2d integration of {0}: {1}s.'.format(os.path.basename(self.img_model.filename), time.time() - t1))
self.cake_img = res[0]
self.cake_tth = res[1]
self.cake_azi = res[2]
return self.cake_img
def create_point_array(self, points, points_ind):
res = []
for i, point_list in enumerate(points):
if point_list.shape == (2,):
res.append([point_list[0], point_list[1], points_ind[i]])
else:
for point in point_list:
res.append([point[0], point[1], points_ind[i]])
return np.array(res)
def get_point_array(self):
return self.create_point_array(self.points, self.points_index)
def get_calibration_parameter(self):
pyFAI_parameter = self.pattern_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
try:
fit2d_parameter = self.pattern_geometry.getFit2D()
fit2d_parameter['polarization_factor'] = self.polarization_factor
except TypeError:
fit2d_parameter = None
pyFAI_parameter['wavelength'] = self.pattern_geometry.wavelength
if fit2d_parameter:
fit2d_parameter['wavelength'] = self.pattern_geometry.wavelength
return pyFAI_parameter, fit2d_parameter
def calculate_number_of_pattern_points(self, max_dist_factor=1.5):
# calculates the number of points for an integrated pattern, based on the distance of the beam center to the the
# image corners. Maximum value is determined by the shape of the image.
fit2d_parameter = self.pattern_geometry.getFit2D()
center_x = fit2d_parameter['centerX']
center_y = fit2d_parameter['centerY']
width, height = self.img_model.img_data.shape
if width > center_x > 0:
side1 = np.max([abs(width - center_x), center_x])
else:
side1 = width
if center_y < height and center_y > 0:
side2 = np.max([abs(height - center_y), center_y])
else:
side2 = height
max_dist = np.sqrt(side1 ** 2 + side2 ** 2)
return int(max_dist * max_dist_factor)
def load(self, filename):
"""
Loads a calibration file and and sets all the calibration parameter.
:param filename: filename for a *.poni calibration file
"""
self.pattern_geometry = AzimuthalIntegrator()
self.pattern_geometry.load(filename)
self.orig_pixel1 = self.pattern_geometry.pixel1
self.orig_pixel2 = self.pattern_geometry.pixel2
self.calibration_name = get_base_name(filename)
self.filename = filename
self.is_calibrated = True
self.create_cake_geometry()
self.set_supersampling()
def save(self, filename):
"""
Saves the current calibration parameters into a a text file. Default extension is
*.poni
"""
self.cake_geometry.save(filename)
self.calibration_name = get_base_name(filename)
self.filename = filename
def create_file_header(self):
try:
# pyFAI version 0.12.0
return self.pattern_geometry.makeHeaders(polarization_factor=self.polarization_factor)
except AttributeError:
# pyFAI after version 0.12.0
from pyFAI.io import DefaultAiWriter
return DefaultAiWriter(None, self.pattern_geometry).make_headers()
def set_fit2d(self, fit2d_parameter):
"""
Reads in a dictionary with fit2d parameters where the fields of the dictionary are:
'directDist', 'centerX', 'centerY', 'tilt', 'tiltPlanRotation', 'pixelX', pixelY',
'polarization_factor', 'wavelength'
"""
self.pattern_geometry.setFit2D(directDist=fit2d_parameter['directDist'],
centerX=fit2d_parameter['centerX'],
centerY=fit2d_parameter['centerY'],
tilt=fit2d_parameter['tilt'],
tiltPlanRotation=fit2d_parameter['tiltPlanRotation'],
pixelX=fit2d_parameter['pixelX'],
pixelY=fit2d_parameter['pixelY'])
self.pattern_geometry.wavelength = fit2d_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = fit2d_parameter['polarization_factor']
self.orig_pixel1 = fit2d_parameter['pixelX'] * 1e-6
self.orig_pixel2 = fit2d_parameter['pixelY'] * 1e-6
self.is_calibrated = True
self.set_supersampling()
def set_pyFAI(self, pyFAI_parameter):
"""
Reads in a dictionary with pyFAI parameters where the fields of dictionary are:
'dist', 'poni1', 'poni2', 'rot1', 'rot2', 'rot3', 'pixel1', 'pixel2', 'wavelength',
'polarization_factor'
"""
self.pattern_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.pattern_geometry.wavelength = pyFAI_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = pyFAI_parameter['polarization_factor']
self.orig_pixel1 = pyFAI_parameter['pixel1']
self.orig_pixel2 = pyFAI_parameter['pixel2']
self.is_calibrated = True
self.set_supersampling()
def load_distortion(self, spline_filename):
self.distortion_spline_filename = spline_filename
self.pattern_geometry.set_splineFile(spline_filename)
if self.cake_geometry:
self.cake_geometry.set_splineFile(spline_filename)
def reset_distortion_correction(self):
self.distortion_spline_filename = None
self.pattern_geometry.set_splineFile(None)
if self.cake_geometry:
self.cake_geometry.set_splineFile(None)
def set_supersampling(self, factor=None):
"""
Sets the supersampling to a specific factor. Whereby the factor determines in how many artificial pixel the
original pixel is split. (factor^2)
factor n_pixel
1 1
2 4
3 9
4 16
"""
if factor is None:
factor = self.supersampling_factor
self.pattern_geometry.pixel1 = self.orig_pixel1 / float(factor)
self.pattern_geometry.pixel2 = self.orig_pixel2 / float(factor)
if factor != self.supersampling_factor:
self.pattern_geometry.reset()
self.supersampling_factor = factor
def reset_supersampling(self):
self.pattern_geometry.pixel1 = self.orig_pixel1
self.pattern_geometry.pixel2 = self.orig_pixel2
def get_two_theta_img(self, x, y):
"""
Gives the two_theta value for the x,y coordinates on the image. Be aware that this function will be incorrect
for pixel indices, since it does not correct for center of the pixel.
:param x: x-coordinate in pixel on the image
:type x: ndarray
:param y: y-coordinate in pixel on the image
:type y: ndarray
:return : two theta in radians
"""
x *= self.supersampling_factor
y *= self.supersampling_factor
return self.pattern_geometry.tth(x - 0.5, y - 0.5)[0] # deletes 0.5 because tth function uses pixel indices
def get_azi_img(self, x, y):
"""
Gives chi for position on image.
:param x: x-coordinate in pixel on the image
:type x: ndarray
:param y: y-coordinate in pixel on the image
:type y: ndarray
:return : azimuth in radians
"""
x *= self.supersampling_factor
y *= self.supersampling_factor
return self.pattern_geometry.chi(x - 0.5, y - 0.5)[0]
def get_two_theta_array(self):
return self.pattern_geometry.twoThetaArray(self.img_model.img_data.shape)[::self.supersampling_factor,
::self.supersampling_factor]
def get_pixel_ind(self, tth, azi):
"""
Calculates pixel index for a specfic two theta and azimutal value.
:param tth:
two theta in radians
:param azi:
azimuth in radians
:return:
tuple of index 1 and 2
"""
tth_ind = find_contours(self.pattern_geometry.ttha, tth)
if len(tth_ind) == 0:
return []
tth_ind = np.vstack(tth_ind)
azi_values = self.pattern_geometry.chi(tth_ind[:, 0], tth_ind[:, 1])
min_index = np.argmin(np.abs(azi_values - azi))
return tth_ind[min_index, 0], tth_ind[min_index, 1]
@property
def wavelength(self):
return self.pattern_geometry.wavelength
class CalibrationModel(object):
def __init__(self, img_model=None):
"""
:param img_model:
:type img_model: ImgModel
"""
self.img_model = img_model
self.points = []
self.points_index = []
self.spectrum_geometry = AzimuthalIntegrator()
self.cake_geometry = None
self.calibrant = Calibrant()
self.start_values = {'dist': 200e-3,
'wavelength': 0.3344e-10,
'pixel_width': 79e-6,
'pixel_height': 79e-6,
'polarization_factor': 0.99}
self.orig_pixel1 = 79e-6
self.orig_pixel2 = 79e-6
self.fit_wavelength = False
self.fit_distance = True
self.is_calibrated = False
self.use_mask = False
self.filename = ''
self.calibration_name = 'None'
self.polarization_factor = 0.99
self.supersampling_factor = 1
self._calibrants_working_dir = os.path.dirname(calibrants.__file__)
self.cake_img = np.zeros((2048, 2048))
self.tth = np.linspace(0, 25)
self.int = np.sin(self.tth)
self.num_points = len(self.int)
self.peak_search_algorithm = None
def find_peaks_automatic(self, x, y, peak_ind):
"""
Searches peaks by using the Massif algorithm
:param x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param y:
y-coordinate in pixel - should be from original image (not supersampled y-coordinate)
:param peak_ind:
peak/ring index to which the found points will be added
:return:
array of points found
"""
massif = Massif(self.img_model._img_data)
cur_peak_points = massif.find_peaks([x, y], stdout=DummyStdOut())
if len(cur_peak_points):
self.points.append(np.array(cur_peak_points))
self.points_index.append(peak_ind)
return np.array(cur_peak_points)
def find_peak(self, x, y, search_size, peak_ind):
"""
Searches a peak around the x,y position. It just searches for the maximum value in a specific search size.
:param x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param y:
y-coordinate in pixel - should be form original image (not supersampled y-coordinate)
:param search_size:
the length of the search rectangle in pixels in all direction in which the algorithm searches for
the maximum peak
:param peak_ind:
peak/ring index to which the found points will be added
:return:
point found (as array)
"""
left_ind = np.round(x - search_size * 0.5)
if left_ind < 0:
left_ind = 0
top_ind = np.round(y - search_size * 0.5)
if top_ind < 0:
top_ind = 0
search_array = self.img_model.img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)]
x_ind, y_ind = np.where(search_array == search_array.max())
x_ind = x_ind[0] + left_ind
y_ind = y_ind[0] + top_ind
self.points.append(np.array([x_ind, y_ind]))
self.points_index.append(peak_ind)
return np.array([np.array((x_ind, y_ind))])
def clear_peaks(self):
self.points = []
self.points_index = []
def create_cake_geometry(self):
self.cake_geometry = AzimuthalIntegrator()
pyFAI_parameter = self.spectrum_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
pyFAI_parameter['wavelength'] = self.spectrum_geometry.wavelength
self.cake_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.cake_geometry.wavelength = pyFAI_parameter['wavelength']
def setup_peak_search_algorithm(self, algorithm, mask=None):
"""
Initializes the peak search algorithm on the current image
:param algorithm:
peak search algorithm used. Possible algorithms are 'Massif' and 'Blob'
:param mask:
if a mask is used during the process this is provided here as a 2d array for the image.
"""
if algorithm == 'Massif':
self.peak_search_algorithm = Massif(self.img_model.get_raw_img_data())
elif algorithm == 'Blob':
if mask is not None:
self.peak_search_algorithm = BlobDetection(self.img_model.get_raw_img_data() * mask)
else:
self.peak_search_algorithm = BlobDetection(self.img_model.get_raw_img_data())
self.peak_search_algorithm.process()
else:
return
def search_peaks_on_ring(self, ring_index, delta_tth=0.1, min_mean_factor=1,
upper_limit=55000, mask=None):
"""
This function is searching for peaks on an expected ring. It needs an initial calibration
before. Then it will search for the ring within some delta_tth and other parameters to get
peaks from the calibrant.
:param ring_index: the index of the ring for the search
:param delta_tth: search space around the expected position in two theta
:param min_mean_factor: a factor determining the minimum peak intensity to be picked up. it is based
on the mean value of the search area defined by delta_tth. Pick a large value
for larger minimum value and lower for lower minimum value. Therefore, a smaller
number is more prone to picking up noise. typical values like between 1 and 3.
:param upper_limit: maximum intensity for the peaks to be picked
:param mask: in case the image has to be masked from certain areas, it need to be given here. Default is None.
The mask should be given as an 2d array with the same dimensions as the image, where 1 denotes a
masked pixel and all others should be 0.
"""
self.reset_supersampling()
if not self.is_calibrated:
return
# transform delta from degree into radians
delta_tth = delta_tth / 180.0 * np.pi
# get appropriate two theta value for the ring number
tth_calibrant_list = self.calibrant.get_2th()
tth_calibrant = np.float(tth_calibrant_list[ring_index])
# get the calculated two theta values for the whole image
if self.spectrum_geometry._ttha is None:
tth_array = self.spectrum_geometry.twoThetaArray(self.img_model._img_data.shape)
else:
tth_array = self.spectrum_geometry._ttha
# create mask based on two_theta position
ring_mask = abs(tth_array - tth_calibrant) <= delta_tth
if mask is not None:
mask = np.logical_and(ring_mask, np.logical_not(mask))
else:
mask = ring_mask
# calculate the mean and standard deviation of this area
sub_data = np.array(self.img_model._img_data.ravel()[np.where(mask.ravel())], dtype=np.float64)
sub_data[np.where(sub_data > upper_limit)] = np.NaN
mean = np.nanmean(sub_data)
std = np.nanstd(sub_data)
# set the threshold into the mask (don't detect very low intensity peaks)
threshold = min_mean_factor * mean + std
mask2 = np.logical_and(self.img_model._img_data > threshold, mask)
mask2[np.where(self.img_model._img_data > upper_limit)] = False
size2 = mask2.sum(dtype=int)
keep = int(np.ceil(np.sqrt(size2)))
try:
sys.stdout = DummyStdOut
res = self.peak_search_algorithm.peaks_from_area(mask2, Imin=mean - std, keep=keep)
sys.stdout = sys.__stdout__
except IndexError:
res = []
# Store the result
if len(res):
self.points.append(np.array(res))
self.points_index.append(ring_index)
self.set_supersampling()
self.spectrum_geometry.reset()
def set_calibrant(self, filename):
self.calibrant = Calibrant()
self.calibrant.load_file(filename)
self.spectrum_geometry.calibrant = self.calibrant
def set_start_values(self, start_values):
self.start_values = start_values
self.polarization_factor = start_values['polarization_factor']
def calibrate(self):
self.spectrum_geometry = GeometryRefinement(self.create_point_array(self.points, self.points_index),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'],
calibrant=self.calibrant)
self.orig_pixel1 = self.start_values['pixel_width']
self.orig_pixel2 = self.start_values['pixel_height']
self.refine()
self.create_cake_geometry()
self.is_calibrated = True
self.calibration_name = 'current'
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.spectrum_geometry.reset()
def refine(self):
self.reset_supersampling()
self.spectrum_geometry.data = self.create_point_array(self.points, self.points_index)
fix = ['wavelength']
if self.fit_wavelength:
fix = []
if not self.fit_distance:
fix.append('dist')
if self.fit_wavelength:
self.spectrum_geometry.refine2()
self.spectrum_geometry.refine2_wavelength(fix=fix)
self.create_cake_geometry()
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.spectrum_geometry.reset()
def integrate_1d(self, num_points=None, mask=None, polarization_factor=None, filename=None,
unit='2th_deg', method='csr'):
if np.sum(mask) == self.img_model.img_data.shape[0] * self.img_model.img_data.shape[1]:
# do not perform integration if the image is completely masked...
return self.tth, self.int
if self.spectrum_geometry._polarization is not None:
if self.img_model.img_data.shape != self.spectrum_geometry._polarization.shape:
# resetting the integrator if the polarization correction matrix has not the correct shape
self.spectrum_geometry.reset()
if polarization_factor is None:
polarization_factor = self.polarization_factor
if num_points is None:
num_points = self.calculate_number_of_spectrum_points(2)
self.num_points = num_points
t1 = time.time()
if unit is 'd_A':
try:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_model.img_data, num_points,
method=method,
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
except NameError:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_model.img_data, num_points,
method='csr',
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
self.tth = self.spectrum_geometry.wavelength / (2 * np.sin(self.tth / 360 * np.pi)) * 1e10
self.int = self.int
else:
try:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_model.img_data, num_points,
method=method,
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
except NameError:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_model.img_data, num_points,
method='csr',
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
logger.info('1d integration of {0}: {1}s.'.format(os.path.basename(self.img_model.filename), time.time() - t1))
ind = np.where((self.int > 0) & (~np.isnan(self.int)))
self.tth = self.tth[ind]
self.int = self.int[ind]
return self.tth, self.int
def integrate_2d(self, mask=None, polarization_factor=None, unit='2th_deg', method='csr', dimensions=(2048, 2048)):
if polarization_factor is None:
polarization_factor = self.polarization_factor
if self.cake_geometry._polarization is not None:
if self.img_model.img_data.shape != self.cake_geometry._polarization.shape:
# resetting the integrator if the polarization correction matrix has not the same shape as the image
self.cake_geometry.reset()
t1 = time.time()
res = self.cake_geometry.integrate2d(self.img_model._img_data, dimensions[0], dimensions[1], method=method,
mask=mask,
unit=unit, polarization_factor=polarization_factor)
logger.info('2d integration of {0}: {1}s.'.format(os.path.basename(self.img_model.filename), time.time() - t1))
self.cake_img = res[0]
self.cake_tth = res[1]
self.cake_azi = res[2]
return self.cake_img
def create_point_array(self, points, points_ind):
res = []
for i, point_list in enumerate(points):
if point_list.shape == (2,):
res.append([point_list[0], point_list[1], points_ind[i]])
else:
for point in point_list:
res.append([point[0], point[1], points_ind[i]])
return np.array(res)
def get_point_array(self):
return self.create_point_array(self.points, self.points_index)
def get_calibration_parameter(self):
pyFAI_parameter = self.cake_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
try:
fit2d_parameter = self.cake_geometry.getFit2D()
fit2d_parameter['polarization_factor'] = self.polarization_factor
except TypeError:
fit2d_parameter = None
try:
pyFAI_parameter['wavelength'] = self.spectrum_geometry.wavelength
fit2d_parameter['wavelength'] = self.spectrum_geometry.wavelength
except RuntimeWarning:
pyFAI_parameter['wavelength'] = 0
return pyFAI_parameter, fit2d_parameter
def calculate_number_of_spectrum_points(self, max_dist_factor=1.5):
# calculates the number of points for an integrated spectrum, based on the distance of the beam center to the the
# image corners. Maximum value is determined by the shape of the image.
fit2d_parameter = self.spectrum_geometry.getFit2D()
center_x = fit2d_parameter['centerX']
center_y = fit2d_parameter['centerY']
width, height = self.img_model.img_data.shape
if center_x < width and center_x > 0:
side1 = np.max([abs(width - center_x), center_x])
else:
side1 = width
if center_y < height and center_y > 0:
side2 = np.max([abs(height - center_y), center_y])
else:
side2 = height
max_dist = np.sqrt(side1 ** 2 + side2 ** 2)
return int(max_dist * max_dist_factor)
def load(self, filename):
"""
Loads a calibration file and and sets all the calibration parameter.
:param filename: filename for a *.poni calibration file
"""
self.spectrum_geometry = AzimuthalIntegrator()
self.spectrum_geometry.load(filename)
self.orig_pixel1 = self.spectrum_geometry.pixel1
self.orig_pixel2 = self.spectrum_geometry.pixel2
self.calibration_name = get_base_name(filename)
self.filename = filename
self.is_calibrated = True
self.create_cake_geometry()
self.set_supersampling()
def save(self, filename):
"""
Saves the current calibration parameters into a a text file. Default extension is
*.poni
"""
self.cake_geometry.save(filename)
self.calibration_name = get_base_name(filename)
self.filename = filename
def create_file_header(self):
return self.cake_geometry.makeHeaders(polarization_factor=self.polarization_factor)
def set_fit2d(self, fit2d_parameter):
"""
Reads in a dictionary with fit2d parameters where the fields of the dictionary are:
'directDist', 'centerX', 'centerY', 'tilt', 'tiltPlanRotation', 'pixelX', pixelY',
'polarization_factor', 'wavelength'
"""
self.spectrum_geometry.setFit2D(directDist=fit2d_parameter['directDist'],
centerX=fit2d_parameter['centerX'],
centerY=fit2d_parameter['centerY'],
tilt=fit2d_parameter['tilt'],
tiltPlanRotation=fit2d_parameter['tiltPlanRotation'],
pixelX=fit2d_parameter['pixelX'],
pixelY=fit2d_parameter['pixelY'])
self.spectrum_geometry.wavelength = fit2d_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = fit2d_parameter['polarization_factor']
self.orig_pixel1 = fit2d_parameter['pixelX'] * 1e-6
self.orig_pixel2 = fit2d_parameter['pixelY'] * 1e-6
self.is_calibrated = True
self.set_supersampling()
def set_pyFAI(self, pyFAI_parameter):
"""
Reads in a dictionary with pyFAI parameters where the fields of dictionary are:
'dist', 'poni1', 'poni2', 'rot1', 'rot2', 'rot3', 'pixel1', 'pixel2', 'wavelength',
'polarization_factor'
"""
self.spectrum_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.spectrum_geometry.wavelength = pyFAI_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = pyFAI_parameter['polarization_factor']
self.orig_pixel1 = pyFAI_parameter['pixel1']
self.orig_pixel2 = pyFAI_parameter['pixel2']
self.is_calibrated = True
self.set_supersampling()
def set_supersampling(self, factor=None):
"""
Sets the supersampling to a specific factor. Whereby the factor determines in how many artificial pixel the
original pixel is split. (factor^2)
factor n_pixel
1 1
2 4
3 9
4 16
"""
if factor is None:
factor = self.supersampling_factor
self.spectrum_geometry.pixel1 = self.orig_pixel1 / float(factor)
self.spectrum_geometry.pixel2 = self.orig_pixel2 / float(factor)
if factor != self.supersampling_factor:
self.spectrum_geometry.reset()
self.supersampling_factor = factor
def reset_supersampling(self):
self.spectrum_geometry.pixel1 = self.orig_pixel1
self.spectrum_geometry.pixel2 = self.orig_pixel2
def get_two_theta_img(self, x, y):
"""
Gives the two_theta value for the x,y coordinates on the image. Be aware that this function will be incorrect
for pixel indices, since it does not correct for center of the pixel.
:return:
two theta in radians
"""
x = np.array([x]) * self.supersampling_factor
y = np.array([y]) * self.supersampling_factor
return self.spectrum_geometry.tth(x - 0.5, y - 0.5)[0] # deletes 0.5 because tth function uses pixel indices
def get_azi_img(self, x, y):
"""
Gives chi for position on image.
:param x:
x-coordinate in pixel
:param y:
y-coordinate in pixel
:return:
azimuth in radians
"""
x *= self.supersampling_factor
y *= self.supersampling_factor
return self.spectrum_geometry.chi(x, y)[0]
def get_two_theta_cake(self, x):
"""
Gives the two_theta value for the x coordinate in the cake
:param x:
x-coordinate on image
:return:
two theta in degree
"""
x -= 0.5
cake_step = self.cake_tth[1] - self.cake_tth[0]
tth = self.cake_tth[int(np.floor(x))] + (x - np.floor(x)) * cake_step
return tth
def get_azi_cake(self, x):
"""
Gives the azimuth value for a cake.
:param x:
x-coordinate in pixel
:return:
azimuth in degree
"""
x -= 0.5
azi_step = self.cake_azi[1] - self.cake_azi[0]
azi = self.cake_azi[int(np.floor(x))] + (x - np.floor(x)) * azi_step
return azi
def get_two_theta_array(self):
return self.spectrum_geometry._ttha[::self.supersampling_factor, ::self.supersampling_factor]
def get_pixel_ind(self, tth, azi):
"""
Calculates pixel index for a specfic two theta and azimutal value.
:param tth:
two theta in radians
:param azi:
azimuth in radians
:return:
tuple of index 1 and 2
"""
tth_ind = find_contours(self.spectrum_geometry.ttha, tth)
tth_ind = np.vstack(tth_ind)
azi_values = self.spectrum_geometry.chi(tth_ind[:, 0], tth_ind[:, 1])
min_index = np.argmin(np.abs(azi_values - azi))
return tth_ind[min_index, 0], tth_ind[min_index, 1]
@property
def wavelength(self):
return self.spectrum_geometry.wavelength
class CalibrationData(object):
def __init__(self, img_data=None):
self.img_data = img_data
self.points = []
self.points_index = []
self.spectrum_geometry = AzimuthalIntegrator()
self.calibrant = Calibrant()
self.start_values = {'dist': 200e-3,
'wavelength': 0.3344e-10,
'pixel_width': 79e-6,
'pixel_height': 79e-6,
'polarization_factor': 0.99}
self.orig_pixel1 = 79e-6
self.orig_pixel2 = 79e-6
self.fit_wavelength = False
self.fit_distance = True
self.is_calibrated = False
self.use_mask = False
self.filename = ''
self.calibration_name = 'None'
self.polarization_factor = 0.99
self.supersampling_factor = 1
self._calibrants_working_dir = os.path.dirname(Calibrants.__file__)
self.cake_img = np.zeros((2048, 2048))
self.tth = np.linspace(0, 25)
self.int = np.sin(self.tth)
def find_peaks_automatic(self, x, y, peak_ind):
"""
Searches peaks by using the Massif algorithm
:param x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param y:
y-coordinate in pixel - should be from original image (not supersampled y-coordinate)
:param peak_ind:
peak/ring index to which the found points will be added
:return:
array of points found
"""
massif = Massif(self.img_data._img_data)
cur_peak_points = massif.find_peaks([x, y])
if len(cur_peak_points):
self.points.append(np.array(cur_peak_points))
self.points_index.append(peak_ind)
return np.array(cur_peak_points)
def find_peak(self, x, y, search_size, peak_ind):
"""
Searches a peak around the x,y position. It just searches for the maximum value in a specific search size.
:param x:
x-coordinate in pixel - should be from original image (not supersampled x-coordinate)
:param y:
y-coordinate in pixel - should be form original image (not supersampled y-coordinate)
:param search_size:
the amount of pixels in all direction in which the algorithm searches for the maximum peak
:param peak_ind:
peak/ring index to which the found points will be added
:return:
point found (as array)
"""
left_ind = np.round(x - search_size * 0.5)
top_ind = np.round(y - search_size * 0.5)
x_ind, y_ind = np.where(self.img_data._img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)] == \
self.img_data._img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)].max())
x_ind = x_ind[0] + left_ind
y_ind = y_ind[0] + top_ind
self.points.append(np.array([x_ind, y_ind]))
self.points_index.append(peak_ind)
return np.array([np.array((x_ind, y_ind))])
def clear_peaks(self):
self.points = []
self.points_index = []
def create_cake_geometry(self):
self.cake_geometry = AzimuthalIntegrator()
pyFAI_parameter = self.spectrum_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
pyFAI_parameter['wavelength'] = self.spectrum_geometry.wavelength
self.cake_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.cake_geometry.wavelength = pyFAI_parameter['wavelength']
def setup_peak_search_algorithm(self, algorithm, mask=None):
"""
Initializes the peak search algorithm on the current image
:param algorithm:
peak search algorithm used. Possible algorithms are 'Massif' and 'Blob'
:param mask:
if a mask is used during the process this is provided here as a 2d array for the image.
"""
if algorithm == 'Massif':
self.peak_search_algorithm = Massif(self.img_data._img_data)
elif algorithm == 'Blob':
if mask is not None:
self.peak_search_algorithm = BlobDetection(self.img_data._img_data * mask)
else:
self.peak_search_algorithm = BlobDetection(self.img_data._img_data)
self.peak_search_algorithm.process()
else:
return
def search_peaks_on_ring(self, peak_index, delta_tth=0.1, min_mean_factor=1,
upper_limit=55000, mask=None):
self.reset_supersampling()
if not self.is_calibrated:
return
# transform delta from degree into radians
delta_tth = delta_tth / 180.0 * np.pi
# get appropriate two theta value for the ring number
tth_calibrant_list = self.calibrant.get_2th()
tth_calibrant = np.float(tth_calibrant_list[peak_index])
# get the calculated two theta values for the whole image
if self.spectrum_geometry._ttha is None:
tth_array = self.spectrum_geometry.twoThetaArray(self.img_data._img_data.shape)
else:
tth_array = self.spectrum_geometry._ttha
# create mask based on two_theta position
ring_mask = abs(tth_array - tth_calibrant) <= delta_tth
if mask is not None:
mask = np.logical_and(ring_mask, np.logical_not(mask))
else:
mask = ring_mask
# calculate the mean and standard deviation of this area
sub_data = np.array(self.img_data._img_data.ravel()[np.where(mask.ravel())], dtype=np.float64)
sub_data[np.where(sub_data > upper_limit)] = np.NaN
mean = np.nanmean(sub_data)
std = np.nanstd(sub_data)
# set the threshold into the mask (don't detect very low intensity peaks)
threshold = min_mean_factor * mean + std
mask2 = np.logical_and(self.img_data._img_data > threshold, mask)
mask2[np.where(self.img_data._img_data > upper_limit)] = False
size2 = mask2.sum(dtype=int)
keep = int(np.ceil(np.sqrt(size2)))
try:
res = self.peak_search_algorithm.peaks_from_area(mask2, Imin=mean - std, keep=keep)
except IndexError:
res = []
# Store the result
if len(res):
self.points.append(np.array(res))
self.points_index.append(peak_index)
self.set_supersampling()
self.spectrum_geometry.reset()
def set_calibrant(self, filename):
self.calibrant = Calibrant()
self.calibrant.load_file(filename)
self.spectrum_geometry.calibrant = self.calibrant
def set_start_values(self, start_values):
self.start_values = start_values
self.polarization_factor = start_values['polarization_factor']
def calibrate(self):
self.spectrum_geometry = GeometryRefinement(self.create_point_array(self.points, self.points_index),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'],
calibrant=self.calibrant)
self.orig_pixel1 = self.start_values['pixel_width']
self.orig_pixel2 = self.start_values['pixel_height']
self.refine()
self.create_cake_geometry()
self.is_calibrated = True
self.calibration_name = 'current'
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.spectrum_geometry.reset()
def refine(self):
self.reset_supersampling()
self.spectrum_geometry.data = self.create_point_array(self.points, self.points_index)
fix = ['wavelength']
if self.fit_wavelength:
fix = []
if not self.fit_distance:
fix.append('dist')
if self.fit_wavelength:
self.spectrum_geometry.refine2()
self.spectrum_geometry.refine2_wavelength(fix=fix)
self.create_cake_geometry()
self.set_supersampling()
# reset the integrator (not the geometric parameters)
self.spectrum_geometry.reset()
def integrate_1d(self, num_points=None, mask=None, polarization_factor=None, filename=None,
unit='2th_deg', method='csr'):
if np.sum(mask) == self.img_data.img_data.shape[0] * self.img_data.img_data.shape[1]:
# do not perform integration if the image is completely masked...
return self.tth, self.int
if self.spectrum_geometry._polarization is not None:
if self.img_data.img_data.shape != self.spectrum_geometry._polarization.shape:
# resetting the integrator if the polarization correction matrix has not the correct shape
self.spectrum_geometry.reset()
if polarization_factor is None:
polarization_factor = self.polarization_factor
if num_points is None:
num_points = self.calculate_number_of_spectrum_points(2)
self.num_points = num_points
t1 = time.time()
if unit is 'd_A':
try:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_data.img_data, num_points,
method=method,
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
except NameError:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_data.img_data, num_points,
method=method,
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
self.tth = self.spectrum_geometry.wavelength / (2 * np.sin(self.tth / 360 * np.pi)) * 1e10
self.int = self.int
else:
try:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_data.img_data, num_points,
method=method,
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
except NameError:
self.tth, self.int = self.spectrum_geometry.integrate1d(self.img_data.img_data, num_points,
method='lut',
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
logger.info('1d integration of {}: {}s.'.format(os.path.basename(self.img_data.filename), time.time() - t1))
ind = np.where((self.int > 0) & (~np.isnan(self.int)))
self.tth = self.tth[ind]
self.int = self.int[ind]
return self.tth, self.int
def integrate_2d(self, mask=None, polarization_factor=None, unit='2th_deg', method='csr', dimensions=(2048, 2048)):
if polarization_factor is None:
polarization_factor = self.polarization_factor
if self.cake_geometry._polarization is not None:
if self.img_data.img_data.shape != self.cake_geometry._polarization.shape:
# resetting the integrator if the polarization correction matrix has not the same shape as the image
self.cake_geometry.reset()
t1 = time.time()
res = self.cake_geometry.integrate2d(self.img_data._img_data, dimensions[0], dimensions[1], method=method,
mask=mask,
unit=unit, polarization_factor=polarization_factor)
logger.info('2d integration of {}: {}s.'.format(os.path.basename(self.img_data.filename), time.time() - t1))
self.cake_img = res[0]
self.cake_tth = res[1]
self.cake_azi = res[2]
return self.cake_img
def create_point_array(self, points, points_ind):
res = []
for i, point_list in enumerate(points):
if point_list.shape == (2,):
res.append([point_list[0], point_list[1], points_ind[i]])
else:
for point in point_list:
res.append([point[0], point[1], points_ind[i]])
return np.array(res)
def get_point_array(self):
return self.create_point_array(self.points, self.points_index)
def get_calibration_parameter(self):
pyFAI_parameter = self.cake_geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
try:
fit2d_parameter = self.cake_geometry.getFit2D()
fit2d_parameter['polarization_factor'] = self.polarization_factor
except TypeError:
fit2d_parameter = None
try:
pyFAI_parameter['wavelength'] = self.spectrum_geometry.wavelength
fit2d_parameter['wavelength'] = self.spectrum_geometry.wavelength
except RuntimeWarning:
pyFAI_parameter['wavelength'] = 0
return pyFAI_parameter, fit2d_parameter
def calculate_number_of_spectrum_points(self, max_dist_factor=1.5):
# calculates the number of points for an integrated spectrum, based on the distance of the beam center to the the
#image corners. Maximum value is determined by the shape of the image.
fit2d_parameter = self.spectrum_geometry.getFit2D()
center_x = fit2d_parameter['centerX']
center_y = fit2d_parameter['centerY']
width, height = self.img_data.img_data.shape
if center_x < width and center_x > 0:
side1 = np.max([abs(width - center_x), center_x])
else:
side1 = width
if center_y < height and center_y > 0:
side2 = np.max([abs(height - center_y), center_y])
else:
side2 = height
max_dist = np.sqrt(side1 ** 2 + side2 ** 2)
return int(max_dist * max_dist_factor)
def load(self, filename):
self.spectrum_geometry = AzimuthalIntegrator()
self.spectrum_geometry.load(filename)
self.orig_pixel1 = self.spectrum_geometry.pixel1
self.orig_pixel2 = self.spectrum_geometry.pixel2
self.calibration_name = get_base_name(filename)
self.filename = filename
self.is_calibrated = True
self.create_cake_geometry()
self.set_supersampling()
def save(self, filename):
self.cake_geometry.save(filename)
self.calibration_name = get_base_name(filename)
self.filename = filename
def create_file_header(self):
return self.cake_geometry.makeHeaders(polarization_factor=self.polarization_factor)
def set_fit2d(self, fit2d_parameter):
print fit2d_parameter
self.spectrum_geometry.setFit2D(directDist=fit2d_parameter['directDist'],
centerX=fit2d_parameter['centerX'],
centerY=fit2d_parameter['centerY'],
tilt=fit2d_parameter['tilt'],
tiltPlanRotation=fit2d_parameter['tiltPlanRotation'],
pixelX=fit2d_parameter['pixelX'],
pixelY=fit2d_parameter['pixelY'])
self.spectrum_geometry.wavelength = fit2d_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = fit2d_parameter['polarization_factor']
self.orig_pixel1 = fit2d_parameter['pixelX'] * 1e-6
self.orig_pixel2 = fit2d_parameter['pixelY'] * 1e-6
self.is_calibrated = True
self.set_supersampling()
def set_pyFAI(self, pyFAI_parameter):
self.spectrum_geometry.setPyFAI(dist=pyFAI_parameter['dist'],
poni1=pyFAI_parameter['poni1'],
poni2=pyFAI_parameter['poni2'],
rot1=pyFAI_parameter['rot1'],
rot2=pyFAI_parameter['rot2'],
rot3=pyFAI_parameter['rot3'],
pixel1=pyFAI_parameter['pixel1'],
pixel2=pyFAI_parameter['pixel2'])
self.spectrum_geometry.wavelength = pyFAI_parameter['wavelength']
self.create_cake_geometry()
self.polarization_factor = pyFAI_parameter['polarization_factor']
self.orig_pixel1 = pyFAI_parameter['pixel1']
self.orig_pixel2 = pyFAI_parameter['pixel2']
self.is_calibrated = True
self.set_supersampling()
def set_supersampling(self, factor=None):
if factor is None:
factor = self.supersampling_factor
self.spectrum_geometry.pixel1 = self.orig_pixel1 / float(factor)
self.spectrum_geometry.pixel2 = self.orig_pixel2 / float(factor)
if factor != self.supersampling_factor:
self.spectrum_geometry.reset()
self.supersampling_factor = factor
def reset_supersampling(self):
self.spectrum_geometry.pixel1 = self.orig_pixel1
self.spectrum_geometry.pixel2 = self.orig_pixel2
def get_two_theta_img(self, x, y):
"""
Gives the two_theta value for the x,y coordinates on the image
:return:
two theta in radians
"""
x = np.array([x]) * self.supersampling_factor
y = np.array([y]) * self.supersampling_factor
return self.spectrum_geometry.tth(x, y)[0]
def get_azi_img(self, x, y):
"""
Gives chi for position on image.
:param x:
x-coordinate in pixel
:param y:
y-coordinate in pixel
:return:
azimuth in radians
"""
x *= self.supersampling_factor
y *= self.supersampling_factor
return self.spectrum_geometry.chi(x, y)[0]
def get_two_theta_cake(self, y):
"""
Gives the two_theta value for the x coordinate in the cake
:param x:
y-coordinate on image
:return:
two theta in degree
"""
return self.cake_tth[np.round(y[0])]
def get_azi_cake(self, x):
"""
Gives the azimuth value for a cake.
:param x:
x-coordinate in pixel
:return:
azimuth in degree
"""
return self.cake_azi[np.round(x[0])]
def get_two_theta_array(self):
return self.spectrum_geometry._ttha[::self.supersampling_factor, ::self.supersampling_factor]
@property
def wavelength(self):
return self.spectrum_geometry.wavelength
__author__ = 'Clemens Prescher'
from pyFAI.blob_detection import BlobDetection
from Data.ImgData import ImgData
import numpy as np
import pylab
img_data = ImgData()
# img_data.load('/Users/Doomgoroth/Programming/Large Projects/Dioptas/Testing/pyFAITest/17_LaB6_dc300-00000.tif')
img_data.load('/Users/Doomgoroth/Programming/Large Projects/Dioptas/Testing/pyFAITest/LaB6_WOS_30keV_005.tif')
bd = BlobDetection(np.log1p(img_data.get_img_data()))
bd.process()
x = []
y = []
int = []
sigma = []
print bd.keypoints.__len__()
for j in range(bd.keypoints.__len__()):
k = bd.keypoints[j]
int.append(k[2])
sigma.append(k[3])
if sigma[-1] > 0.25:
x.append(k[0])
y.append(k[1])
pylab.hist(int)
pylab.figure(2)
pylab.hist(sigma)
class CalibrationData(object):
def __init__(self, img_data=None):
self.img_data = img_data
self.points = []
self.points_index = []
self.geometry = AzimuthalIntegrator()
self.calibrant = Calibrant()
self.start_values = {
'dist': 200e-3,
'wavelength': 0.3344e-10,
'pixel_width': 79e-6,
'pixel_height': 79e-6,
'polarization_factor': 0.95
}
self.fit_wavelength = False
self.is_calibrated = False
self.use_mask = False
self.calibration_name = 'None'
self.polarization_factor = 0.95
self._calibrants_working_dir = os.path.dirname(Calibrants.__file__)
def find_peaks_automatic(self, x, y, peak_ind):
massif = Massif(self.img_data.img_data)
cur_peak_points = massif.find_peaks([x, y])
if len(cur_peak_points):
self.points.append(np.array(cur_peak_points))
self.points_index.append(peak_ind)
return np.array(cur_peak_points)
def find_peak(self, x, y, search_size, peak_ind):
left_ind = np.round(x - search_size * 0.5)
top_ind = np.round(y - search_size * 0.5)
x_ind, y_ind = np.where(self.img_data.img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)] == \
self.img_data.img_data[left_ind:(left_ind + search_size),
top_ind:(top_ind + search_size)].max())
x_ind = x_ind[0] + left_ind
y_ind = y_ind[0] + top_ind
self.points.append(np.array([x_ind, y_ind]))
self.points_index.append(peak_ind)
return np.array([np.array((x_ind, y_ind))])
def clear_peaks(self):
self.points = []
self.points_index = []
def setup_peak_search_algorithm(self, algorithm, mask=None):
# init the peak search algorithm
if algorithm == 'Massif':
self.peak_search_algorithm = Massif(self.img_data.img_data)
elif algorithm == 'Blob':
if mask is not None:
self.peak_search_algorithm = BlobDetection(
self.img_data.img_data * mask)
else:
self.peak_search_algorithm = BlobDetection(
self.img_data.img_data)
self.peak_search_algorithm.process()
else:
return
def search_peaks_on_ring(self,
peak_index,
delta_tth=0.1,
min_mean_factor=1,
upper_limit=55000,
mask=None):
if not self.is_calibrated:
return
#transform delta from degree into radians
delta_tth = delta_tth / 180.0 * np.pi
# get appropiate two theta value for the ring number
tth_calibrant_list = self.calibrant.get_2th()
tth_calibrant = np.float(tth_calibrant_list[peak_index])
# get the calculated two theta values for the whole image
if self.geometry._ttha is None:
tth_array = self.geometry.twoThetaArray(
self.img_data.img_data.shape)
else:
tth_array = self.geometry._ttha
# create mask based on two_theta position
ring_mask = abs(tth_array - tth_calibrant) <= delta_tth
if mask is not None:
mask = np.logical_and(ring_mask, np.logical_not(mask))
else:
mask = ring_mask
# calculate the mean and standard deviation of this area
sub_data = np.array(self.img_data.img_data.ravel()[np.where(
mask.ravel())],
dtype=np.float64)
sub_data[np.where(sub_data > upper_limit)] = np.NaN
mean = np.nanmean(sub_data)
std = np.nanstd(sub_data)
# set the threshold into the mask (don't detect very low intensity peaks)
threshold = min_mean_factor * mean + std
mask2 = np.logical_and(self.img_data.img_data > threshold, mask)
mask2[np.where(self.img_data.img_data > upper_limit)] = False
size2 = mask2.sum(dtype=int)
keep = int(np.ceil(np.sqrt(size2)))
try:
res = self.peak_search_algorithm.peaks_from_area(mask2,
Imin=mean - std,
keep=keep)
except IndexError:
res = []
# Store the result
if len(res):
self.points.append(np.array(res))
self.points_index.append(peak_index)
def set_calibrant(self, filename):
self.calibrant = Calibrant()
self.calibrant.load_file(filename)
self.geometry.calibrant = self.calibrant
def set_start_values(self, start_values):
self.start_values = start_values
self.polarization_factor = start_values['polarization_factor']
def calibrate(self):
self.geometry = GeometryRefinement(
self.create_point_array(self.points, self.points_index),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'],
calibrant=self.calibrant)
self.refine()
self.integrate()
self.is_calibrated = True
self.calibration_name = 'current'
def refine(self):
self.geometry.data = self.create_point_array(self.points,
self.points_index)
self.geometry.refine2()
if self.fit_wavelength:
self.geometry.refine2_wavelength(fix=[])
def integrate(self):
self.integrate_1d()
self.integrate_2d()
def integrate_1d(self,
num_points=1400,
mask=None,
polarization_factor=None,
filename=None,
unit='2th_deg'):
if np.sum(
mask
) == self.img_data.img_data.shape[0] * self.img_data.img_data.shape[1]:
#do not perform integration if the image is completelye masked...
return self.tth, self.int
if polarization_factor is None:
polarization_factor = self.polarization_factor
if unit is 'd_A':
self.tth, self.int = self.geometry.integrate1d(
self.img_data.img_data,
num_points,
method='lut',
unit='2th_deg',
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
ind = np.where(self.tth > 0)
self.tth = self.geometry.wavelength / (
2 * np.sin(self.tth[ind] / 360 * np.pi)) * 1e10
self.int = self.int[ind]
else:
self.tth, self.int = self.geometry.integrate1d(
self.img_data.img_data,
num_points,
method='lut',
unit=unit,
mask=mask,
polarization_factor=polarization_factor,
filename=filename)
if self.int.max() > 0:
ind = np.where(self.int > 0)
self.tth = self.tth[ind]
self.int = self.int[ind]
return self.tth, self.int
def integrate_2d(self,
mask=None,
polarization_factor=None,
unit='2th_deg'):
if polarization_factor is None:
polarization_factor = self.polarization_factor
res = self.geometry.integrate2d(
self.img_data.img_data,
2048,
2048,
method='lut',
mask=mask,
unit=unit,
polarization_factor=polarization_factor)
self.cake_img = res[0]
self.cake_tth = res[1]
self.cake_azi = res[2]
return self.cake_img
def create_point_array(self, points, points_ind):
res = []
for i, point_list in enumerate(points):
if point_list.shape == (2, ):
res.append([point_list[0], point_list[1], points_ind[i]])
else:
for point in point_list:
res.append([point[0], point[1], points_ind[i]])
return np.array(res)
def get_point_array(self):
return self.create_point_array(self.points, self.points_index)
def get_calibration_parameter(self):
pyFAI_parameter = self.geometry.getPyFAI()
pyFAI_parameter['polarization_factor'] = self.polarization_factor
try:
fit2d_parameter = self.geometry.getFit2D()
fit2d_parameter['polarization_factor'] = self.polarization_factor
except TypeError:
fit2d_parameter = None
try:
pyFAI_parameter['wavelength'] = self.geometry.wavelength
fit2d_parameter['wavelength'] = self.geometry.wavelength
except RuntimeWarning:
pyFAI_parameter['wavelength'] = 0
return pyFAI_parameter, fit2d_parameter
def load(self, filename):
self.geometry = GeometryRefinement(
np.zeros((2, 3)),
dist=self.start_values['dist'],
wavelength=self.start_values['wavelength'],
pixel1=self.start_values['pixel_width'],
pixel2=self.start_values['pixel_height'])
self.geometry.load(filename)
self.calibration_name = get_base_name(filename)
self.is_calibrated = True
def save(self, filename):
self.geometry.save(filename)
self.calibration_name = get_base_name(filename)