def smoothing(self, sigma_zero):
"""
:param sigma_zero: initial sigma as a first smoothing
:return:void (images are saved at disc space in path+ '/npy_arrays_3DGaussianFiltering'
"""
path_to_save = '/3DGaussianSmoothing/'
sigmas_x = (self.k**self.powers) * sigma_zero
sigmas_y = (self.k**self.powers) * sigma_zero
sigmas_z = ((self.k**self.powers) * sigma_zero) / (self.spacing[2] /
self.spacing[0])
print(sigmas_x, sigmas_z)
saving = SaveImage(self.path + path_to_save)
for sigma_x, sigma_y, sigma_z in zip(sigmas_x, sigmas_y, sigmas_z):
im = self.image.Image3D
image = normalize(im, [np.min(im), np.max(im)], [-1.0, 1.0])
print('befor ', np.min(image), np.max(image))
smoothed_image = gaussian_filter(image,
(sigma_x, sigma_y, sigma_z))
smoothed_image = smoothed_image.astype(dtype=np.float32)
smoothed_image = normalize(smoothed_image, [-1.0, 1.0], [0.0, 1.0])
temp_image = deepcopy(self.image)
temp_image.Image3D = smoothed_image
temp_image.sigma = sigma_x
saving.saveImage(temp_image)
def smoothing(self, sigma_zero):
"""
:param sigma_zero: initial sigma as a first smoothing
:return:void (images are saved at disc space in path+ '/npy_arrays_2DGaussianFiltering'
"""
path_to_save = '/2DGaussianSmoothing/'
sigmas_x = (self.k**self.powers) * sigma_zero / self.spacing[0]
sigmas_y = (self.k**self.powers) * sigma_zero / self.spacing[1]
# make directory
im = self.image.Image3D[:, :, 20]
saving = SaveImage(self.path + path_to_save)
for sigma_x, sigma_y in zip(sigmas_x, sigmas_y):
image = normalize(im, [np.min(im), np.max(im)], [-1.0, 1.0])
smoothed_image = gaussian_filter(image, (sigma_x, sigma_y))
print(np.min(smoothed_image), np.max(smoothed_image))
smoothed_image = normalize(smoothed_image, [-1.0, 1.0], [0.0, 1.0])
temp_image = deepcopy(self.image)
temp_image.Image3D = smoothed_image
temp_image.sigma = sigma_x
saving.saveImage(temp_image)
def smoothing(self, sigma_zero):
"""
:param sigma_zero: initial sigma as a first smoothing
:return:void (images are saved at disc space in path+ '/npy_arrays_3DGaussianFiltering'
"""
path_to_save = '/3DGaussianSmoothing/'
sigmas_x = (self.k ** self.powers) * sigma_zero
sigmas_y = (self.k ** self.powers) * sigma_zero
sigmas_z = ((self.k ** self.powers) * sigma_zero) / (self.spacing[2] / self.spacing[0])
print(sigmas_x, sigmas_z)
saving = SaveImage(self.path + path_to_save)
for sigma_x, sigma_y, sigma_z in zip(sigmas_x, sigmas_y, sigmas_z):
im = self.image.Image3D
image = normalize(im, [np.min(im), np.max(im)], [-1.0, 1.0])
print('befor ', np.min(image), np.max(image))
smoothed_image = gaussian_filter(image, (sigma_x, sigma_y, sigma_z))
smoothed_image = smoothed_image.astype(dtype=np.float32)
smoothed_image = normalize(smoothed_image, [-1.0, 1.0], [0.0, 1.0])
temp_image = deepcopy(self.image)
temp_image.Image3D = smoothed_image
temp_image.sigma = sigma_x
saving.saveImage(temp_image)
def smoothing(self, sigma_zero):
"""
:param sigma_zero: initial sigma as a first smoothing
:return:void (images are saved at disc space in path+ '/npy_arrays_2DGaussianFiltering'
"""
path_to_save = '/2DGaussianSmoothing/'
sigmas_x = (self.k ** self.powers) * sigma_zero / self.spacing[0]
sigmas_y = (self.k ** self.powers) * sigma_zero / self.spacing[1]
# make directory
im = self.image.Image3D[:, :, 20]
saving = SaveImage(self.path + path_to_save)
for sigma_x, sigma_y in zip(sigmas_x, sigmas_y):
image = normalize(im, [np.min(im), np.max(im)], [-1.0, 1.0])
smoothed_image = gaussian_filter(image, (sigma_x, sigma_y))
print(np.min(smoothed_image), np.max(smoothed_image))
smoothed_image = normalize(smoothed_image, [-1.0, 1.0], [0.0, 1.0])
temp_image = deepcopy(self.image)
temp_image.Image3D = smoothed_image
temp_image.sigma = sigma_x
saving.saveImage(temp_image)
def get_Histogram2D(self, elevation, azimuth, weights, plt=None):
'''
:param elevation: array with elevations
:param azimuth: array with azimuth
:param weights: array with weights
:return: histogram of occurrence
'''
H, e, z = np.histogram2d(elevation.flatten(), azimuth.flatten(), bins=[self.No_bin_x, self.No_bin_y],
normed=False, weights=weights.flatten(), range=[[0.0, np.pi], [0.0, 2 * np.pi]])
H = normalize(H, [np.min(H), np.max(H)], [0., 1.])
H = normalize(H, [np.min(H), np.max(H)], [-1., 1.])
H = gaussian_filter(H, 0.1, truncate=3.0)
H = normalize(H, [np.min(H), np.max(H)], [0., 1.])
return H
def get_Histogram2D(self, elevation, azimuth, weights, plt=None):
'''
:param elevation: array with elevations
:param azimuth: array with azimuth
:param weights: array with weights
:return: histogram of occurrence
'''
H, e, z = np.histogram2d(elevation.flatten(),
azimuth.flatten(),
bins=[self.No_bin_x, self.No_bin_y],
normed=False,
weights=weights.flatten(),
range=[[0.0, np.pi], [0.0, 2 * np.pi]])
H = normalize(H, [np.min(H), np.max(H)], [0., 1.])
H = normalize(H, [np.min(H), np.max(H)], [-1., 1.])
H = gaussian_filter(H, 0.1, truncate=3.0)
H = normalize(H, [np.min(H), np.max(H)], [0., 1.])
return H
def apply(self):
list_of_image = self.ReadImage.openImage()
saving = SaveImage(self.path + self.path_to_save)
import numpy as np
for i in range(0, len(list_of_image) - 1):
DoG = list_of_image[i + 1].Image3D - list_of_image[i].Image3D
print('before', np.min(DoG), np.max(DoG))
DoG = normalize(DoG, [-1, 1], [0, 1])
print(DoG.dtype)
print(np.min(DoG), np.max(DoG))
list_of_image[i].Image3D = DoG
saving.saveImage(list_of_image[i])
def keypoints_histograms(self, image3D_agregator, keypooints_agregator):
# diff in [mm space]
dx, dy, dz = np.gradient(image3D_agregator.Image3D, image3D_agregator.spacing[0], image3D_agregator.spacing[1],
image3D_agregator.spacing[2])
keypoints = keypoints_concatenate(keypooints_agregator)
# do konfigracji
delta_azimuth = np.pi / 4.
delta_elevation = np.pi / 4.
#solid_azimuth=np.arange(0,2*np.pi+delta_elevation,delta_elevation)
#solid_angle = 1.0 / (delta_elevation * (np.cos(solid_azimuth) - np.cos(solid_azimuth + delta_azimuth)))
#solid_angle = normalize(solid_angle, [np.min(solid_angle), np.max(solid_angle)], [0, 1])
keypoint_list = []
for k in range(0, keypoints.shape[0]):
try:
i = keypoints[k][0]
j = keypoints[k][1]
z = keypoints[k][2]
except IndexError:
keypoint_list.append([0,0,0,0,0])
pass
continue
temp_x = dx[i - self.size_of_window_x:i + self.size_of_window_x + 1,
j - self.size_of_window_x:j + self.size_of_window_x + 1,
z - self.size_of_window_z:z + self.size_of_window_z + 1]
temp_y = dy[i - self.size_of_window_x:i + self.size_of_window_x + 1,
j - self.size_of_window_x:j + self.size_of_window_x + 1,
z - self.size_of_window_z:z + self.size_of_window_z + 1]
temp_z = dz[i - self.size_of_window_x:i + self.size_of_window_x + 1,
j - self.size_of_window_x:j + self.size_of_window_x + 1,
z - self.size_of_window_z:z + self.size_of_window_z + 1]
if temp_x.shape[0] != 2 * self.size_in_pixels_xy + 1: continue
if temp_x.shape[1] != 2 * self.size_in_pixels_xy + 1: continue
if temp_x.shape[2] != 2 * self.size_in_pixels_xy + 1: continue
self.magnitude = np.sqrt(temp_x ** 2 + temp_y ** 2 + temp_z ** 2)
self.azimuth = np.arctan2(temp_y, temp_x) + np.pi
self.elevation = np.arctan2(temp_z,np.sqrt(temp_x ** 2 + temp_y ** 2))+np.pi/2
self.magnitude = normalize(self.magnitude, [np.min(self.magnitude), np.max(self.magnitude)], [0, 1])
self.gaussian_weight = normalize(self.gaussian_weight, [np.min(self.gaussian_weight),
np.max(self.gaussian_weight)], [0, 1])
weights = self.magnitude * self.gaussian_weight #* solid_angle
self.weights = normalize(weights, [np.min(weights), np.max(weights)], [0, 1])
NO_xbin = 4
NO_ybin = 8
H2D = Histogram2D(NO_xbin, NO_ybin)
H2D.apply(self.elevation, self.azimuth, weights)
#self.H2D = H2D.get_Histogram2D()
#fig =figure()
from mpl_toolkits.mplot3d import Axes3D
#ax = fig.add_subplot(111,projection='3d')
#barchart(H2D.H)
#mlab.colorbar(nb_labels=2,label_fmt='%.1f')
#show()
angles = H2D.get_Histogram2D_max()
if angles.size == 4:
keypoint_list.append([i, j, z, angles[0][0], angles[0][1]])
keypoint_list.append([i, j, z, angles[1][0], angles[1][1]])
elif angles.size == 2:
keypoint_list.append([i, j, z, angles[0][0], angles[0][1]])
return keypoint_list
def apply(self, mask, spacing):
mask = self.interp(mask, spacing)
# constant for histogram
delta_azimuth = np.pi / 4.
delta_elevation = np.pi / 4.
# mesurmet for histogram
dx, dy, dz = np.gradient(mask)
magnitude = np.sqrt(dx**2 + dy**2 + dz**2)
azimuth = np.arctan2(dy, dx) + np.pi
elevation = np.arctan2(dz, np.sqrt(dx**2 + dy**2)) + np.pi / 2
# weights with normalization for equal contribution
# solid_angle = 1 / (delta_elevation * (np.cos(azimuth) - np.cos(azimuth + delta_azimuth)))
# solid_angle = normalize(solid_angle, [np.min(solid_angle), np.max(solid_angle)], [0, 1])
self.magnitude = normalize(
magnitude,
[np.min(magnitude), np.max(magnitude)], [0, 1])
self.gaussian_weight = normalize(
self.gaussian_weight,
[np.min(self.gaussian_weight),
np.max(self.gaussian_weight)], [0, 1])
weights = magnitude * self.gaussian_weight # * solid_angle
self.weights = normalize(
weights, [np.min(weights), np.max(weights)], [0, 1])
# splitig array mask for 8 equal parts
div0 = np.array([
8,
])
div1 = np.array([
7,
])
# angles
h0 = np.hsplit(azimuth, div0)[0]
A_1 = np.vsplit(h0, div0)[0] # 1
A_2 = np.vsplit(h0, div0)[0] # 2
A_3 = np.vsplit(h0, div1)[1] # 3
A_4 = np.vsplit(h0, div1)[1] #
h1 = np.hsplit(azimuth, div1)[1]
A_5 = np.vsplit(h1, div0)[0] # 5
A_6 = np.vsplit(h1, div0)[0] # 6
A_7 = np.vsplit(h1, div1)[1] # 7
A_8 = np.vsplit(h1, div1)[1] # 8
h0 = np.hsplit(elevation, div0)[0]
E_1 = np.vsplit(h0, div0)[0]
E_2 = np.vsplit(h0, div0)[0]
E_3 = np.vsplit(h0, div1)[1] # 3
E_4 = np.vsplit(h0, div1)[1] #
h1 = np.hsplit(elevation, div1)[1]
E_5 = np.vsplit(h1, div0)[0] # 5
E_6 = np.vsplit(h1, div0)[0] # 6
E_7 = np.vsplit(h1, div1)[1] # 7
E_8 = np.vsplit(h1, div1)[1] # 8
# weights
w0 = np.hsplit(self.weights, div0)[0]
W_1 = np.vsplit(h0, div0)[0] # 1
W_2 = np.vsplit(h0, div0)[0] # 2
W_3 = np.vsplit(h0, div1)[1] # 3
W_4 = np.vsplit(h0, div1)[1] #
w1 = np.hsplit(self.weights, div1)[1]
W_5 = np.vsplit(h1, div0)[0] # 5
W_6 = np.vsplit(h1, div0)[0] # 6
W_7 = np.vsplit(h1, div1)[1] # 7
W_8 = np.vsplit(h1, div1)[1] # 8
# constatn for histogram
NO_xbin = 4
NO_ybin = 8
H2D = Histogram2D(NO_xbin, NO_ybin)
list_of_histograms_for_keypoint = []
for i in range(1, 9):
nr = str(i)
list_of_histograms_for_keypoint.append(
H2D.get_Histogram2D(eval('E_' + nr), eval('A_' + nr),
eval('W_' + nr)))
# list of histograms for descriptor (8 histograms as 4x8 bins)
return np.array(list_of_histograms_for_keypoint)
def apply(self, mask, spacing):
mask = self.interp(mask, spacing)
# constant for histogram
delta_azimuth = np.pi / 4.
delta_elevation = np.pi / 4.
# mesurmet for histogram
dx, dy, dz = np.gradient(mask)
magnitude = np.sqrt(dx ** 2 + dy ** 2 + dz ** 2)
azimuth = np.arctan2(dy, dx) + np.pi
elevation = np.arctan2(dz, np.sqrt(dx ** 2 + dy ** 2)) + np.pi / 2
# weights with normalization for equal contribution
# solid_angle = 1 / (delta_elevation * (np.cos(azimuth) - np.cos(azimuth + delta_azimuth)))
# solid_angle = normalize(solid_angle, [np.min(solid_angle), np.max(solid_angle)], [0, 1])
self.magnitude = normalize(magnitude, [np.min(magnitude), np.max(magnitude)], [0, 1])
self.gaussian_weight = normalize(self.gaussian_weight, [np.min(self.gaussian_weight),
np.max(self.gaussian_weight)], [0, 1])
weights = magnitude * self.gaussian_weight # * solid_angle
self.weights = normalize(weights, [np.min(weights), np.max(weights)], [0, 1])
# splitig array mask for 8 equal parts
div0 = np.array([8, ])
div1 = np.array([7, ])
# angles
h0 = np.hsplit(azimuth, div0)[0]
A_1 = np.vsplit(h0, div0)[0] # 1
A_2 = np.vsplit(h0, div0)[0] # 2
A_3 = np.vsplit(h0, div1)[1] # 3
A_4 = np.vsplit(h0, div1)[1] #
h1 = np.hsplit(azimuth, div1)[1]
A_5 = np.vsplit(h1, div0)[0] # 5
A_6 = np.vsplit(h1, div0)[0] # 6
A_7 = np.vsplit(h1, div1)[1] # 7
A_8 = np.vsplit(h1, div1)[1] # 8
h0 = np.hsplit(elevation, div0)[0]
E_1 = np.vsplit(h0, div0)[0]
E_2 = np.vsplit(h0, div0)[0]
E_3 = np.vsplit(h0, div1)[1] # 3
E_4 = np.vsplit(h0, div1)[1] #
h1 = np.hsplit(elevation, div1)[1]
E_5 = np.vsplit(h1, div0)[0] # 5
E_6 = np.vsplit(h1, div0)[0] # 6
E_7 = np.vsplit(h1, div1)[1] # 7
E_8 = np.vsplit(h1, div1)[1] # 8
# weights
w0 = np.hsplit(self.weights, div0)[0]
W_1 = np.vsplit(h0, div0)[0] # 1
W_2 = np.vsplit(h0, div0)[0] # 2
W_3 = np.vsplit(h0, div1)[1] # 3
W_4 = np.vsplit(h0, div1)[1] #
w1 = np.hsplit(self.weights, div1)[1]
W_5 = np.vsplit(h1, div0)[0] # 5
W_6 = np.vsplit(h1, div0)[0] # 6
W_7 = np.vsplit(h1, div1)[1] # 7
W_8 = np.vsplit(h1, div1)[1] # 8
# constatn for histogram
NO_xbin = 4
NO_ybin = 8
H2D = Histogram2D(NO_xbin, NO_ybin)
list_of_histograms_for_keypoint = []
for i in range(1, 9):
nr = str(i)
list_of_histograms_for_keypoint.append(
H2D.get_Histogram2D(eval('E_' + nr), eval('A_' + nr), eval('W_' + nr)))
# list of histograms for descriptor (8 histograms as 4x8 bins)
return np.array(list_of_histograms_for_keypoint)