def updateImage(self, val):
"""
Hlavni update metoda.
Cinny kod pro gaussovske filtrovani, prahovani, binarni uzavreni a otevreni a vraceni nejvetsich nebo oznacenych objektu.
"""
if (sys.version_info[0] < 3):
import copy
self.imgFiltering = copy.copy(self.data)
else:
self.imgFiltering = self.data.copy()
## Filtrovani
## Zjisteni jakou sigmu pouzit
if(self.firstRun == True and self.inputSigma >= 0):
sigma = numpy.round(self.inputSigma, 2)
else:
sigma = numpy.round(self.ssigma.val, 2)
sigmaNew = thresholding_functions.calculateSigma(self.voxel, sigma)
self.imgFiltering = thresholding_functions.gaussFilter(self.imgFiltering, sigmaNew)
del(sigmaNew)
## Prahovani (smin, smax)
max_threshold = -1
min_threshold = self.threshold
if self.interactivity:
self.smin.val = (numpy.round(self.smin.val, 2))
self.smin.valtext.set_text('{}'.format(self.smin.val))
self.smax.val = (numpy.round(self.smax.val, 2))
self.smax.valtext.set_text('{}'.format(self.smax.val))
min_threshold = self.smin.val
max_threshold = self.smax.val
self.threshold = min_threshold
if (self.threshold == -1) and self.firstRun:
min_threshold = thresholding_functions.calculateAutomaticThreshold(self.imgFiltering, self.arrSeed)
self.imgFiltering = thresholding_functions.thresholding(self.imgFiltering, min_threshold, max_threshold, True, self.interactivity)
## Operace binarni otevreni a uzavreni.
## Nastaveni hodnot slideru.
if (self.interactivity == True) :
closeNum = int(numpy.round(self.sclose.val, 0))
openNum = int(numpy.round(self.sopen.val, 0))
self.sclose.valtext.set_text('{}'.format(closeNum))
self.sopen.valtext.set_text('{}'.format(openNum))
else:
closeNum = self.ICBinaryClosingIterations
openNum = self.ICBinaryOpeningIterations
self.imgFiltering = thresholding_functions.binaryClosingOpening(self.imgFiltering, closeNum, openNum, True)
## Zjisteni nejvetsich objektu.
self.getBiggestObjects()
## Vykresleni dat
if (self.interactivity == True):
self.drawVisualization()
## Nastaveni kontrolnich hodnot
self.firstRun = False
garbage.collect()
self.debugInfo()
def generate_sample_data(m=1, noise_level=0.005, gauss_sigma=0.1):
"""
Generate sample vessel system.
J. Kunes
| Input:
| m - output will be (100*m)^3 numpy array
| noise_level - noise power, disable noise with -1
| gauss_sigma - gauss filter sigma, disable filter with -1
| Output:
| (100*m)^3 numpy array
| voxel size = [1,1,1]
"""
import thresholding_functions
data3d = np.zeros((100 * m, 100 * m, 100 * m), dtype=np.int)
# size 8
data3d_new = np.ones((100 * m, 100 * m, 100 * m), dtype=np.bool)
data3d_new[0:30 * m, 20 * m, 20 * m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 8 * m] = 0
data3d[data3d_new == 0] = 1
# size 7
data3d_new = np.ones((100 * m, 100 * m, 100 * m), dtype=np.bool)
data3d_new[31 * m:70 * m, 20 * m, 20 * m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 7 * m] = 0
data3d[data3d_new == 0] = 1
# size 6
data3d_new = np.ones((100 * m, 100 * m, 100 * m), dtype=np.bool)
data3d_new[70 * m, 20 * m:50 * m, 20 * m] = 0
data3d_new[31 * m, 20 * m, 20 * m:70 * m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 6 * m] = 0
data3d[data3d_new == 0] = 1
# size 5
data3d_new = np.ones((100 * m, 100 * m, 100 * m), dtype=np.bool)
data3d_new[70 * m:95 * m, 20 * m, 20 * m] = 0
data3d_new[31 * m:60 * m, 20 * m, 70 * m] = 0
data3d_new[70 * m:90 * m, 50 * m, 20 * m] = 0
data3d_new[70 * m, 50 * m, 20 * m:50 * m] = 0
data3d_new[31 * m, 20 * m: 45 * m, 20 * m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 5*m] = 0
data3d[data3d_new == 0] = 1
# size 4
data3d_new = np.ones((100*m, 100*m, 100*m), dtype=np.bool)
data3d_new[31*m, 20*m:50*m, 70*m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 4*m] = 0
data3d[data3d_new == 0] = 1
# size 3
data3d_new = np.ones((100*m, 100*m, 100*m), dtype=np.bool)
data3d_new[31*m:50*m, 50*m, 70*m] = 0
data3d_new[31*m:50*m, 45*m, 20*m] = 0
data3d_new[70*m, 50*m:70*m, 50*m] = 0
data3d_new[70*m:80*m, 50*m, 50*m] = 0
data3d_new[scipy.ndimage.distance_transform_edt(data3d_new) <= 3*m] = 0
data3d[data3d_new == 0] = 1
data3d = data3d*3030 # 3030+5920 = vessel value
data3d += 5920 # 5920 = background value
if gauss_sigma > 0:
sigma = np.round(gauss_sigma, 2)
sigmaNew = thresholding_functions.calculateSigma([1, 1, 1], sigma)
data3d = thresholding_functions.gaussFilter(data3d, sigmaNew)
if noise_level > 0:
noise = np.random.normal(1, noise_level, (100*m, 100*m, 100*m))
data3d = data3d*noise
return data3d