def LabelContacts(mydict, imageSize):
mykeys = mydict.keys()
nLink = len(mykeys)
memoryType = utilities.BestMemoryType(nLink)
labeledLinks = np.zeros(imageSize, dtype=memoryType)
for iLink in range(nLink):
ind = mydict[mykeys[iLink]]
X, Y, Z = np.unravel_index(ind, imageSize)
labeledLinks[X, Y, Z] = iLink + 1
return labeledLinks
def PoresSegmentation(myImg,
phases={'void': False},
structuringElement=np.ones((3, 3, 3)),
seedMethod='hMaxima',
seedParam=4,
distanceType='ITKDanielson'):
#Phases=dict, phases['myphase']=codeForMyPhaseInImage
pores = np.zeros(myImg.shape, dtype=np.uint8)
watershedLines = np.zeros(myImg.shape, dtype=np.bool)
if distanceType == 'euclidean' or distanceType == 'ITKDanielson':
memoryType = np.float16
elif distanceType == 'chamfer':
memoryType = np.int8
distanceMap = np.zeros(myImg.shape, dtype=memoryType)
labelShift = [0]
for phaseName in phases.keys():
phaseCode = phases[phaseName]
phaseImage = (myImg == np.uint8(phaseCode)).astype(np.bool)
poresPhase, watershedLinesPhase, distanceMapPhase = PoresWatershedSegmentationOnePhase(
phaseImage,
structuringElement=structuringElement,
seedMethod=seedMethod,
seedParam=seedParam,
distanceType=distanceType)
phaseImage = phaseImage.astype(np.bool)
memoryType = utilities.BestMemoryType(poresPhase.max() + pores.max())
pores = pores.astype(memoryType)
pores[phaseImage] = (
poresPhase[phaseImage]).astype(memoryType) + labelShift[-1] * (
poresPhase[phaseImage] > 0).astype(memoryType)
watershedLines[phaseImage] = watershedLinesPhase[phaseImage]
distanceMap[phaseImage] = distanceMapPhase[phaseImage]
del phaseImage, poresPhase, watershedLinesPhase, distanceMapPhase
labelShift.append(pores.max())
#phaseBoundaries = SobelEdgeDetection(myImg)
phaseBoundaries = imageAnalysis.FeatureExtraction.SobelEdgeDetection(myImg)
watershedLines = np.logical_or(watershedLines, phaseBoundaries)
porePhase = np.zeros(pores.max(), dtype=np.uint8)
for i in range(len(labelShift) - 1):
porePhase[np.arange(labelShift[i],
labelShift[i + 1])] = int(phases[phases.keys()[i]])
return pores, watershedLines, distanceMap, porePhase
def __ReadInputs__(image, phases, seedMethod, seedParam):
if isinstance(image, basestring):
imageFileName = image
assert (os.path.isfile(imageFileName))
image = utilities.ReadTiff(imageFileName)
else:
assert (isinstance(image, np.ndarray))
assert (image.min() >= 0)
memoryType = utilities.BestMemoryType(image.max())
image = image.astype(memoryType)
assert (isinstance(phases, dict))
assert (isinstance(int(seedParam), int))
assert (seedMethod == "hMaxima" or seedMethod == "localMax")
return image, phases, seedMethod, seedParam
def MedianFilter(image,radius = (3,3,3)):
"""MedianFilter : see SimpleITK.Median
:param image : numpy image
:param radius: size of neighborhood used to take median
:return: numpy image after filtering
:Example:
import VirtualMaterials as vmat
filteredImage = vmat.ImageAnalysis.Filters.MedianFilter(image,radius = (3,3,3))
"""
assert(image.min()>=0)
memoryType = utils.BestMemoryType(image.max())
itkimage = sitk.GetImageFromArray(image.astype(memoryType))
itkimage = sitk.Median(itkimage,radius = radius)
image=sitk.GetArrayFromImage(itkimage).astype(memoryType)
return image
def OtsuThreshold(image,
insideValue = 100,
outsideValue = 0,
numberOfHistogramBins = 128,
maskOutput = True,
maskValue = 255 ):
"""OtsuThreshold : see SimpleITK.OtsuThreshold
:param image : numpy image
:param insideValue:
:param outsideValue:
:param numberOfHistogramBins:
:param maskOutput:
:param maskValue:
:return: numpy image after filtering
:Example:
import VirtualMaterials as vmat
filteredImage = vmat.ImageAnalysis.Filters.OtsuThreshold(image,
insideValue = 100,
outsideValue = 0,
numberOfHistogramBins = 128,
maskOutput = True,
maskValue = 255 )
"""
assert(image.min()>=0)
memoryType = memoryType = utils.BestMemoryType(image.max())
itkimage = sitk.GetImageFromArray(image.astype(memoryType))
#OTSU Threshold FILTER
itkimage= sitk.OtsuThreshold(
itkimage,
np.uint8(insideValue),
np.uint8(outsideValue),
np.uint32(numberOfHistogramBins),
bool(maskOutput),
np.uint8(maskValue)
)
image=sitk.GetArrayFromImage(itkimage).astype(memoryType)
return image
def Threshold(image, lower=0.0,upper=1.0,outsideValue=0.0):
"""Threshold : see SimpleITK.Threshold
:param image : numpy image
:param lower:
:param upper:
:param outsideValue:
:return: numpy image after filtering
:Example:
import VirtualMaterials as vmat
filteredImage = vmat.ImageAnalysis.Filters.Threshold(image, lower=0.0,upper=1.0,outsideValue=0.0)
"""
assert(image.min()>=0)
memoryType = utils.BestMemoryType(image.max())
itkimage = sitk.GetImageFromArray(image.astype(memoryType))
itkimage = sitk.Threshold(itkimage,
lower=lower,
upper=upper,
outsideValue=outsideValue)
image=sitk.GetArrayFromImage(itkimage).astype(memoryType)
return image
def PoresWatershedSegmentationOnePhase(phaseImage,
structuringElement=np.ones((3, 3, 3)),
distanceType='ITKDanielson',
watershedAlgo='ITK',
seedMethod='hMaxima',
seedParam=4):
#Calcul de la carte de distance distanceMap
if distanceType == 'euclidean':
memoryType = np.float16
distanceMap = ndimage.distance_transform_edt(phaseImage).astype(
memoryType)
elif distanceType == 'chamfer':
memoryType = np.int8
distanceMap = ndimage.distance_transform_cdt(
phaseImage, metric='chessboard').astype(memoryType)
elif distanceType == 'ITKDanielson':
memoryType = np.float16
itkimage = sitk.GetImageFromArray(
np.logical_not(phaseImage).astype(np.uint8))
itkdistanceMap = sitk.DanielssonDistanceMap(itkimage)
del itkimage
distanceMap = sitk.GetArrayFromImage(itkdistanceMap).astype(memoryType)
del itkdistanceMap
#Choix des marqueurs pour la segmentation (centres des pores) : H-maxima :
#maxima de la carte de distance dont les pics sont étêtés d'une hauteur h. Utilise
#une recontruction morphologique pour construire la carte de distance étêtée.
hMaximaLib = 'skimage'
if seedMethod == 'localMax':
seedParam = int(seedParam)
local_maxi = feature.peak_local_max(distanceMap.astype(np.float),
min_distance=seedParam,
indices=False)
elif seedMethod == 'hMaxima' and hMaximaLib == 'skimage' and seedParam > 0:
hContrast = np.asarray(seedParam).astype(memoryType)
reconstructed = morphology.reconstruction(
distanceMap - hContrast, distanceMap).astype(memoryType)
local_maxi = (distanceMap - reconstructed).astype(np.bool)
del reconstructed
elif seedMethod == 'hMaxima' and hMaximaLib == 'ITK' and seedParam > 0:
hContrast = np.asarray(seedParam).astype(memoryType)
itkSeedimage = sitk.GetImageFromArray(
(distanceMap - hContrast).astype(np.float32))
itkMarkerImage = sitk.GetImageFromArray(distanceMap.astype(np.float32))
itkReconstructed = sitk.ReconstructionByDilation(
itkSeedimage, itkMarkerImage)
del itkMarkerImage, itkSeedimage
reconstructed = sitk.GetArrayFromImage(itkReconstructed).astype(
memoryType)
del itkReconstructed
local_maxi = ((distanceMap - reconstructed) > hContrast / 10).astype(
np.bool)
del reconstructed
markers = ndimage.measurements.label(local_maxi,
structure=structuringElement)[0]
del local_maxi
#Calcul des lignes de partage de niveau
if watershedAlgo == 'Mahotas':
_, watershedLines = mahotas.cwatershed(
(distanceMap.max() - distanceMap).astype(np.int8),
markers,
Bc=structuringElement,
return_lines=True)
elif watershedAlgo == 'ITK':
itkMarkers = sitk.GetImageFromArray(markers)
itkDistance = sitk.GetImageFromArray(distanceMap.astype(np.float))
itkDistance = sitk.InvertIntensity(itkDistance)
wsITK = sitk.MorphologicalWatershedFromMarkers(itkDistance,
itkMarkers,
markWatershedLine=True,
fullyConnected=True)
del itkMarkers, itkDistance
# mask = itk.MaskImageFilter.IUC2IUC2IUC2.New(ws, fill)
# overlay = itk.LabelOverlayImageFilter.IUC2IUC2IRGBUC2.New(reader,
# mask,
# UseBackground=True)
ws = sitk.GetArrayFromImage(wsITK).astype(np.uint8)
del wsITK
watershedLines = (ws == 0).astype(np.bool)
watershedLines[np.logical_not(phaseImage)] = False
del markers
#Labeler des pores séparés par les lignes de partage de niveau
pores = ndimage.measurements.label(np.logical_and(
phaseImage, np.logical_not(watershedLines)),
structure=structuringElement)[0]
pores[np.logical_not(phaseImage)] = 0
memoryType = utilities.BestMemoryType(pores.max())
pores = pores.astype(memoryType)
return pores, watershedLines, distanceMap