Skip to content

GitLab

Commit 045c962a authored by Edward Andò's avatar Edward Andò
Browse files

declassified ITKwatershed

parent b695f90e
Pipeline #62887 failed with stages
in 8 minutes and 31 seconds
Hide whitespace changes
Inline Side-by-side
requirements.txt
View file @ 045c962a
# numpy==1.20.1 # Python >=3.7
numpy==1.19.5 # Python >=3.6
......@@ -33,8 +32,8 @@ pyamg==4.0.0
# mpi4py==3.0.3
# Jupyter
ipykernel==5.5.0 # Python >=3.5
# ipython==7.21.0 # Python >=3.7
#ipykernel==5.5.0 # Python >=3.5
#ipython==7.21.0 # Python >=3.7
ipython==7.16.1 # Python >=3.6
# R (correlated random field)
......@@ -43,9 +42,9 @@ rpy2==3.4.2
#############
# Graphical #
#############
PyQt5==5.15.4 # Python >=3.6
PyQt5-sip==12.8.1 # Python >=3.5
qimage2ndarray==1.8.3
#PyQt5==5.15.4 # Python >=3.6
#PyQt5-sip==12.8.1 # Python >=3.5
#qimage2ndarray==1.8.3
##############
......
tools/label/ITKwatershed.py
View file @ 045c962a
......@@ -16,132 +16,10 @@ You should have received a copy of the GNU General Public License along with
this program. If not, see <http://www.gnu.org/licenses/>.
"""
from __future__ import print_function
import numpy
import tifffile
import sys
import os
import SimpleITK
import spam.label
#if sys.version_info >= (3, 0):
#def xrange(*args, **kwargs):
#return iter(range(*args, **kwargs))
"""
Classes binding ITK morphological watershed algorithm implementation
Note:
Author is Olumide Okubadejo
"""
class _MWatershed(object):
def __init__(self, data, level=1, watershedLineOn=False, fullyConnected=True, fromMarkers=False, markerImage=None):
self._level = level
self._watershedLineOn = watershedLineOn
self._fullyConnected = fullyConnected
self._data = SimpleITK.GetImageFromArray(data)
self._fromMarkers = fromMarkers
if fromMarkers:
self._markerImage = SimpleITK.GetImageFromArray(markerImage)
#def _CalculateGradient(self):
#gradient = SimpleITK.GradientMagnitudeImageFilter()
#self._gradient = gradient.Execute(self._bdata)
def _Binarize(self):
threshold = SimpleITK.OtsuThresholdImageFilter()
self._bdata = threshold.Execute(self._data)
self._bdata = self._Rescale(self._bdata)
self._threshold = threshold.GetThreshold()
def _Rescale(self, data):
rescale = SimpleITK.RescaleIntensityImageFilter()
return rescale.Execute(data, 0, 65535)
def _DistanceMap(self):
distanceMap = SimpleITK.DanielssonDistanceMapImageFilter()
self._distance = distanceMap.Execute(self._bdata)
pass
def _InvertImage(self, data):
invert = SimpleITK.InvertIntensityImageFilter()
inverted = invert.Execute(data)
return inverted
def _MaskImage(self):
mask = SimpleITK.MaskImageFilter()
self._mask = mask.Execute(self._labelImage, self._bdata)
def _FillHoles(self):
fill = SimpleITK.BinaryFillholeImageFilter()
self._bdata = fill.Execute(self._bdata)
def _LabelOverlay(self):
overlay = SimpleITK.LabelOverlayImageFilter()
self._overlay = overlay.Execute(self._bdata, self._mask)
def _Watershed(self):
if (self._fromMarkers):
watershed = SimpleITK.MorphologicalWatershedFromMarkersImageFilter()
else:
watershed = SimpleITK.MorphologicalWatershedImageFilter()
watershed.SetFullyConnected(self._fullyConnected)
watershed.SetMarkWatershedLine(self._watershedLineOn)
if (self._fromMarkers):
self._labelImage = watershed.Execute(
self._distance, self._markerImage)
else:
watershed.SetLevel(self._level)
self._labelImage = watershed.Execute(self._distance)
#def GetThreshold(self):
#return self._threshold
#def GetLabeledImage(self):
#return SimpleITK.GetArrayFromImage(self._mask)
#def GetITKLabeledImage(self):
#return self._labelImage
#def GetColorMap(self):
#return SimpleITK.GetArrayFromImage(self._overlay)
#def GetBinaryImage(self):
#return SimpleITK.GetArrayFromImage(self._bdata)
#def GetDistanceMapImage(self):
#return SimpleITK.GetArrayFromImage(self._distance)
def GetMaskImage(self):
return SimpleITK.GetArrayFromImage(self._mask)
#def GetLabelParams(self):
#return self._labelParams.GetLabels()
#def GetLabelObject(self):
#return self._labelParams
#def GetRadiusArray(self):
#return self._labelParams.GetRadiusArray()
def run(self):
self._Binarize()
# self._bdata = self._InvertImage(self._bdata)
self._FillHoles()
self._DistanceMap()
self._distance = self._InvertImage(self._distance)
self._Watershed()
self._bdata = self._InvertImage(self._bdata)
self._MaskImage()
self._LabelOverlay()
#self._labelParams = _Labels(self._data, self._labelImage)
return SimpleITK.GetArrayFromImage(self._overlay)
def watershed(binary, markers=None, verbose = False):
def watershed(binary, markers=None, watershedLevel=1):
"""
This function runs an ITK watershed on a binary image and returns a labelled image.
This function uses an interpixel watershed.
......@@ -154,40 +32,77 @@ def watershed(binary, markers=None, verbose = False):
markers : 3D numpy array (optional, default = None)
Not implemented yet, but try!
verbose : boolean (optional, default = False)
True for printing the evolution of process
False for not printing the evolution of process
watershedLevel : int, optional
Watershed level for merging maxima
Returns
--------
labelled : 3D numpy array of ints
3D array where each object is numbered
"""
import matplotlib.pyplot as plt
def plotme(im):
im = SimpleITK.GetArrayFromImage(im)
plt.imshow(im[im.shape[0]//2])
plt.show()
binary = binary > 0
# Let's convert it 8-bit
binary = binary.astype('<u1') * 255
bdata = SimpleITK.GetImageFromArray(binary)
if markers is not None:
if verbose:
print("\tITKwatershed.watershed(): Running watershed with your markers...", end='')
if markers.max() > 65535: markers = markers.astype('<u4')
else: markers = markers.astype('<u2')
markers = markers.astype('<u4') if markers.max() > 65535 else markers.astype('<u2')
markers = SimpleITK.GetImageFromArray(markers)
watershedLevel=1
watershedLineOn=False
fullyConnected=True
#thresholdFilt = SimpleITK.OtsuThresholdImageFilter()
#bdata = thresholdFilt.Execute(data)
#rescaleFilt = SimpleITK.RescaleIntensityImageFilter()
#rescaleFilt.SetOutputMinimum(0)
#rescaleFilt.SetOutputMaximum(65535)
#bdata = rescaleFilt.Execute(data)
mWatershed = _MWatershed(binary, markerImage=markers.astype(spam.label.labelType), fromMarkers=True,
level=1, watershedLineOn=False, fullyConnected=True)
#threshold = thresholdFilt.GetThreshold()
#fillFilt = SimpleITK.BinaryFillholeImageFilter()
#bdata = fillFilt.Execute(bdata)
invertFilt = SimpleITK.InvertIntensityImageFilter()
bdata = invertFilt.Execute(bdata)
distanceMapFilt = SimpleITK.DanielssonDistanceMapImageFilter()
distance = distanceMapFilt.Execute(bdata)
distance = invertFilt.Execute(distance)
if markers is None:
watershedFilt = SimpleITK.MorphologicalWatershedImageFilter()
else:
if verbose:
print("\tITKwatershed.watershed(): Running watershed...", end='')
mWatershed = _MWatershed(
binary, level=1, watershedLineOn=False, fullyConnected=True)
labels = mWatershed.run()
if verbose:
print("done.")
if verbose:
print("\tITKwatershed.watershed(): Collecting labelled image...", end='')
lab = mWatershed.GetMaskImage().astype('<u4')
if verbose:
print("done.")
watershedFilt = SimpleITK.MorphologicalWatershedFromMarkersImageFilter()
watershedFilt.SetFullyConnected(fullyConnected)
watershedFilt.SetMarkWatershedLine(watershedLineOn)
if markers is None:
watershedFilt.SetLevel(watershedLevel)
labelImage = watershedFilt.Execute(distance)
else:
labelImage = watershedFilt.Execute(distance, markers)
bdata = invertFilt.Execute(bdata)
maskFilt = SimpleITK.MaskImageFilter()
mask = maskFilt.Execute(labelImage, bdata)
#overlayFilt = SimpleITK.LabelOverlayImageFilter()
#overlay = overlayFilt.Execute(bdata, mask)
lab = SimpleITK.GetArrayFromImage(mask).astype(spam.label.labelType)
return lab
tools/tests/test_label.py
View file @ 045c962a
......@@ -601,7 +601,7 @@ class TestFunctionLabel(unittest.TestCase):
box = numpy.zeros((100, 100, 100), dtype='<f8')
spam.kalisphera.makeSphere(box, [[50, 50, 50-20], [50, 50, 50+20]], [20, 20])
box = box > 0.5
lab = spam.label.watershed(box, verbose=True)
lab = spam.label.watershed(box)
labVolumes = spam.label.volumes(lab)
setVoronoi = spam.label.setVoronoi(lab)
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment