def __test_all_on_image(self, image):
for bt in self.BOUNDARY_TERMS_2ARGS:
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
boundary_term=bt,
boundary_term_args=(image, False))
self.__execute(graph, self.image)
for bt in self.BOUNDARY_TERMS_3ARGS:
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
boundary_term=bt,
boundary_term_args=(image, 1.0, False))
self.__execute(graph, self.image)
def __test_all_on_image(self, image):
for bt in self.BOUNDARY_TERMS_2ARGS:
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
boundary_term=bt,
boundary_term_args=(image, False))
self.__execute(graph, self.image)
for bt in self.BOUNDARY_TERMS_3ARGS:
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
boundary_term=bt,
boundary_term_args=(image, 1.0, False))
self.__execute(graph, self.image)
def test_voxel_based(self):
"""Executes the complete pipeline of the graph cut algorithm."""
# create the graph from the image
original_image = scipy.asarray(self.__voriginal_image)
graph = graph_from_voxels(scipy.asarray(self.__vfg_markers),
scipy.asarray(self.__vbg_markers),
boundary_term=boundary_difference_linear,
boundary_term_args=(original_image, False))
# execute min-cut / executing BK_MFMC
try:
maxflow = graph.maxflow()
except Exception as e:
self.fail('An error was thrown during the external executions: {}'.
format(e.message))
# reshape results to form a valid mask
result = scipy.zeros(original_image.size, dtype=scipy.bool_)
for idx in range(len(result)):
result[idx] = 0 if graph.termtype.SINK == graph.what_segment(
idx) else 1
result = result.reshape(original_image.shape)
# check results for validity
self.assertTrue(
(result == scipy.asarray(self.__vexpected)).all(),
'Resulting voxel-based cut is different than expected.')
self.assertEqual(
maxflow, self.__vmaxflow,
'The resulting maxflow {} differs from the expected one {}.'.
format(maxflow, self.__vmaxflow))
def __test_regional_term_2d(self, term, term_args):
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
regional_term=term,
regional_term_args=term_args)
result = self.__execute(graph, self.image)
assert_array_equal(result, self.result)
def __test_regional_term_2d(self, term, term_args):
graph = graph_from_voxels(self.fgmarkers,
self.bgmarkers,
regional_term=term,
regional_term_args=term_args)
result = self.__execute(graph, self.image)
assert_array_equal(result, self.result)
def graphcut_as_postprocessing(heatmap, image):
fgmarkers = (heatmap > 0.99).astype(np.float)
fgmarkers_ = np.zeros(shape=(*fgmarkers.shape, 3))
for i in range(3):
fgmarkers_[:, :, i] = fgmarkers
bgmarkers = (heatmap < 0.01).astype(np.float)
bgmarkers_ = np.zeros(shape=(*bgmarkers.shape, 3))
for i in range(3):
bgmarkers_[:, :, i] = bgmarkers
image = image.cpu().data.numpy()
image = np.moveaxis(image, 0, 2)
image = np.dot(image[..., :3], [0.299, 0.587, 0.114])
# image = image[:,:,0]
#
# image = np.asarray([[0, 0, 0, 0, 0],
# [0, 0, 2, 0, 0],
# [0, 0, 2, 0, 0],
# [0, 0, 2, 0, 0],
# [0, 0, 2, 0, 0]], dtype=np.float)
# fgmarkers = np.asarray([[0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0],
# [0, 0, 1, 0, 0]], dtype=np.bool)
# bgmarkers = np.asarray([[1, 0, 0, 0, 1],
# [0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0]], dtype=np.bool)
# expected = image.astype(np.bool)
graph = graph_from_voxels(fgmarkers,
bgmarkers,
boundary_term=boundary_difference_exponential,
boundary_term_args=(image, 1000, (10, 10)))
try:
graph.maxflow()
except Exception as e:
print(e)
# reshape results to form a valid mask
result = np.zeros(image.size, dtype=np.bool)
for idx in range(len(result)):
result[idx] = 0 if graph.termtype.SINK == graph.what_segment(
idx) else 1
return (result * 1).reshape(image.shape)
def med_graphcut(img, alpha, sigma):
# to be initialized
'''
:alpha: balancing between boundary and regional term.
:sigma: use in the boundary term. sigma^2 = 1/beta.
: !spacing: A sequence containing the slice spacing used for weighting
the computed neighbourhood weight value for different dimensions
:param img:
:return:
'''
# sigma, alpha = 40. , 1e-2
spacing = False
# fg_thresh, bg_thresh = 200, 1 # values between them are Undetermined.
# set foreground, background seeds
fg_mark, bg_mark = np.zeros_like(img), np.zeros_like(img)
img[img < 0] = 0 # ignore negative intensity
# only for arterial_essence suppression
max_essence_intensity = 320
img[img > max_essence_intensity] = max_essence_intensity
# fg_mark[img>=fg_thresh] = 1
# bg_mark[img<=bg_thresh] = 1
# probability map
# A defect: the histogram should not be the whole image, but the foreground distribution.
# probability_map = prob_map(img, prob_histogram(img, (bg_thresh+1, 255))) # bins=256 or others
# probability_map = np.zeros(img.shape)+1e-4 # only use boundary term
probability_map = prob_map_md(img)
# print(probability_map)
gcgraph = graph_from_voxels(fg_mark,
bg_mark,
regional_term=regional_probability_map,
regional_term_args=(probability_map, alpha),
boundary_term=boundary_difference_exponential,
boundary_term_args=(img, sigma, spacing))
maxflow = gcgraph.maxflow()
result_mask = np.zeros(img.size, dtype=np.bool)
for idx in range(len(result_mask)):
result_mask[idx] = 0 if gcgraph.termtype.SINK == gcgraph.what_segment(
idx) else 1 # ?
result_mask = result_mask.reshape(img.shape)
return result_mask, (str(sigma), str(alpha))
def test_spacing(self):
image = numpy.asarray(
[[0, 0, 0, 0, 0], [0, 0, 2, 0, 0], [0, 0, 2, 0, 0],
[0, 0, 2, 0, 0], [0, 0, 2, 0, 0]],
dtype=numpy.float)
fgmarkers = numpy.asarray(
[[0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [0, 0, 0, 0, 0],
[0, 0, 0, 0, 0], [0, 0, 1, 0, 0]],
dtype=numpy.bool)
bgmarkers = numpy.asarray(
[[1, 0, 0, 0, 1], [0, 0, 0, 0, 0], [0, 0, 0, 0, 0],
[0, 0, 0, 0, 0], [0, 0, 0, 0, 0]],
dtype=numpy.bool)
expected = image.astype(numpy.bool)
graph = graph_from_voxels(fgmarkers,
bgmarkers,
boundary_term=boundary_difference_division,
boundary_term_args=(image, 1.0, (1., 5.0)))
result = self.__execute(graph, image)
assert_array_equal(result, expected)
def test_spacing(self):
image = numpy.asarray([[0,0,0,0,0],
[0,0,2,0,0],
[0,0,2,0,0],
[0,0,2,0,0],
[0,0,2,0,0]], dtype=numpy.float)
fgmarkers = numpy.asarray([[0,0,0,0,0],
[0,0,0,0,0],
[0,0,0,0,0],
[0,0,0,0,0],
[0,0,1,0,0]], dtype=numpy.bool)
bgmarkers = numpy.asarray([[1,0,0,0,1],
[0,0,0,0,0],
[0,0,0,0,0],
[0,0,0,0,0],
[0,0,0,0,0]], dtype=numpy.bool)
expected = image.astype(numpy.bool)
graph = graph_from_voxels(fgmarkers,
bgmarkers,
boundary_term=boundary_difference_division,
boundary_term_args=(image, 1.0, (1., 5.0)))
result = self.__execute(graph, image)
assert_array_equal(result, expected)
def segmentacionGraphCut(imagen,
foreground,
background,
sigma=15.0,
spacing=(1.0, 1.0, 1.0)):
print("Pasamos las imagenes a medpy")
coche_medpy = skimage_a_medpy(imagen)
cocheFore_medpy = foreground.transpose()
cocheBack_medpy = background.transpose()
print("Creamos el grafo")
grafo = graph_from_voxels(
np.array([cocheFore_medpy, cocheFore_medpy,
cocheFore_medpy]), # Marcas de foreground
np.array([cocheBack_medpy, cocheBack_medpy,
cocheBack_medpy]), # Marcas de backgorund
boundary_term=boundary_difference_exponential, #Calculo pesos bordes
boundary_term_args=(coche_medpy, sigma, spacing))
print("Calculamos el maxflow")
grafo.maxflow()
print("Calculmaos el resultado")
resultado_medpy = np.zeros(coche_medpy.size, dtype=np.bool)
for idx in range(len(resultado_medpy)):
if grafo.termtype.SINK == grafo.what_segment(idx):
resultado_medpy[idx] = False
else:
resultado_medpy[idx] = True
resultado_medpy = resultado_medpy.reshape(coche_medpy.shape)
resultado = medpy_a_skimage(resultado_medpy)
mask = np.zeros([resultado.shape[0], resultado.shape[1]], dtype=bool)
mask[np.count_nonzero(resultado == True, axis=2) > 1.5] = True
resultado = medpy_a_skimage(
np.array([mask.transpose(),
mask.transpose(),
mask.transpose()]))
resultado = resultadoMarcado(imagen, resultado * 255)
io.imshow(resultado)
return mask, resultado
def test_voxel_based(self):
"""Executes the complete pipeline of the graph cut algorithm."""
# create the graph from the image
original_image = scipy.asarray(self.__voriginal_image)
graph = graph_from_voxels(scipy.asarray(self.__vfg_markers),
scipy.asarray(self.__vbg_markers),
boundary_term=boundary_difference_linear,
boundary_term_args=original_image)
# execute min-cut / executing BK_MFMC
try:
maxflow = graph.maxflow()
except Exception as e:
self.fail('An error was thrown during the external executions: {}'.format(e.message))
# reshape results to form a valid mask
result = scipy.zeros(original_image.size, dtype=scipy.bool_)
for idx in range(len(result)):
result[idx] = 0 if graph.termtype.SINK == graph.what_segment(idx) else 1
result = result.reshape(original_image.shape)
# check results for validity
self.assertTrue((result == scipy.asarray(self.__vexpected)).all(), 'Resulting voxel-based cut is different than expected.')
self.assertEqual(maxflow, self.__vmaxflow, 'The resulting maxflow {} differs from the expected one {}.'.format(maxflow, self.__vmaxflow))
def main():
# parse cmd arguments
parser = getParser()
parser.parse_args()
args = getArguments(parser)
# prepare logger
logger = Logger.getInstance()
if args.debug: logger.setLevel(logging.DEBUG)
elif args.verbose: logger.setLevel(logging.INFO)
# check if output image exists
if not args.force:
if os.path.exists(args.output):
logger.warning('The output image {} already exists. Exiting.'.format(args.output))
exit(-1)
# select boundary term
['diff_linear', 'diff_exp', 'diff_div', 'diff_pow', 'max_linear', 'max_exp', 'max_div', 'max_pow']
if 'diff_linear' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_linear
logger.info('Selected boundary term: linear difference of intensities')
elif 'diff_exp' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_exponential
logger.info('Selected boundary term: exponential difference of intensities')
elif 'diff_div' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_division
logger.info('Selected boundary term: divided difference of intensities')
elif 'diff_pow' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_power
logger.info('Selected boundary term: power based / raised difference of intensities')
elif 'max_linear' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_linear
logger.info('Selected boundary term: linear maximum of intensities')
elif 'max_exp' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_exponential
logger.info('Selected boundary term: exponential maximum of intensities')
elif 'max_div' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_division
logger.info('Selected boundary term: divided maximum of intensities')
elif 'max_pow' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_power
logger.info('Selected boundary term: power based / raised maximum of intensities')
# load input images
badditional_image_data, reference_header = load(args.badditional)
markers_image_data, _ = load(args.markers)
# split marker image into fg and bg images
fgmarkers_image_data, bgmarkers_image_data = split_marker(markers_image_data)
# check if all images dimensions are the same
if not (badditional_image_data.shape == fgmarkers_image_data.shape == bgmarkers_image_data.shape):
logger.critical('Not all of the supplied images are of the same shape.')
raise ArgumentError('Not all of the supplied images are of the same shape.')
# extract spacing if required
if args.spacing:
spacing = header.get_pixel_spacing(reference_header)
logger.info('Taking spacing of {} into account.'.format(spacing))
else:
spacing = False
# generate graph
logger.info('Preparing BK_MFMC C++ graph...')
gcgraph = graphcut.graph_from_voxels(fgmarkers_image_data,
bgmarkers_image_data,
boundary_term = boundary_term,
boundary_term_args = (badditional_image_data, args.sigma, spacing))
# execute min-cut
logger.info('Executing min-cut...')
maxflow = gcgraph.maxflow()
logger.debug('Maxflow is {}'.format(maxflow))
# reshape results to form a valid mask
logger.info('Applying results...')
result_image_data = scipy.zeros(bgmarkers_image_data.size, dtype=scipy.bool_)
for idx in range(len(result_image_data)):
result_image_data[idx] = 0 if gcgraph.termtype.SINK == gcgraph.what_segment(idx) else 1
result_image_data = result_image_data.reshape(bgmarkers_image_data.shape)
# save resulting mask
save(result_image_data.astype(scipy.bool_), args.output, reference_header, args.force)
logger.info('Successfully terminated.')
def main():
# parse cmd arguments
parser = getParser()
parser.parse_args()
args = getArguments(parser)
# prepare logger
logger = Logger.getInstance()
if args.debug: logger.setLevel(logging.DEBUG)
elif args.verbose: logger.setLevel(logging.INFO)
# check if output image exists
if not args.force:
if os.path.exists(args.output):
logger.warning('The output image {} already exists. Exiting.'.format(args.output))
exit(-1)
# select boundary term
['diff_linear', 'diff_exp', 'diff_div', 'diff_pow', 'max_linear', 'max_exp', 'max_div', 'max_pow']
if 'diff_linear' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_linear
logger.info('Selected boundary term: linear difference of intensities')
elif 'diff_exp' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_exponential
logger.info('Selected boundary term: exponential difference of intensities')
elif 'diff_div' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_division
logger.info('Selected boundary term: divided difference of intensities')
elif 'diff_pow' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_difference_power
logger.info('Selected boundary term: power based / raised difference of intensities')
elif 'max_linear' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_linear
logger.info('Selected boundary term: linear maximum of intensities')
elif 'max_exp' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_exponential
logger.info('Selected boundary term: exponential maximum of intensities')
elif 'max_div' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_division
logger.info('Selected boundary term: divided maximum of intensities')
elif 'max_pow' == args.boundary:
boundary_term = graphcut.energy_voxel.boundary_maximum_power
logger.info('Selected boundary term: power based / raised maximum of intensities')
# load input images
badditional_image_data, reference_header = load(args.badditional)
markers_image_data, _ = load(args.markers)
# split marker image into fg and bg images
fgmarkers_image_data, bgmarkers_image_data = split_marker(markers_image_data)
# check if all images dimensions are the same
if not (badditional_image_data.shape == fgmarkers_image_data.shape == bgmarkers_image_data.shape):
logger.critical('Not all of the supplied images are of the same shape.')
raise ArgumentError('Not all of the supplied images are of the same shape.')
# extract spacing if required
if args.spacing:
spacing = header.get_pixel_spacing(reference_header)
logger.info('Taking spacing of {} into account.'.format(spacing))
else:
spacing = False
# generate graph
logger.info('Preparing BK_MFMC C++ graph...')
gcgraph = graphcut.graph_from_voxels(fgmarkers_image_data,
bgmarkers_image_data,
boundary_term = boundary_term,
boundary_term_args = (badditional_image_data, args.sigma, spacing))
# execute min-cut
logger.info('Executing min-cut...')
maxflow = gcgraph.maxflow()
logger.debug('Maxflow is {}'.format(maxflow))
# reshape results to form a valid mask
logger.info('Applying results...')
result_image_data = scipy.zeros(bgmarkers_image_data.size, dtype=scipy.bool_)
for idx in range(len(result_image_data)):
result_image_data[idx] = 0 if gcgraph.termtype.SINK == gcgraph.what_segment(idx) else 1
result_image_data = result_image_data.reshape(bgmarkers_image_data.shape)
# save resulting mask
save(result_image_data.astype(scipy.bool_), args.output, reference_header, args.force)
logger.info('Successfully terminated.')
