def test_neighbor_midy(self):
"""Test 2."""
im = np.zeros((5, 5))
vals = im_utils.neighbor_vals(im, 0, 2)
# make assertion
assert vals[1] == 0.
assert vals[2] == 0.
assert vals[4] == 0.
assert vals[6] == 0.
assert vals[7] == 0.
def test_neighbor_endymidx(self):
"""Test 8."""
im = np.zeros((5, 5))
vals = im_utils.neighbor_vals(im, 2, 4)
# make assertion
assert vals[0] == 0.
assert vals[1] == 0.
assert vals[2] == 0.
assert vals[3] == 0.
assert vals[4] == 0.
def test_neighbor_endyzero(self):
"""Test 7."""
im = np.zeros((5, 5))
vals = im_utils.neighbor_vals(im, 2, 0)
# make assertion
assert vals[3] == 0.
assert vals[4] == 0.
assert vals[5] == 0.
assert vals[6] == 0.
assert vals[7] == 0.
def test_neighbor_endxmid(self):
"""Test 6."""
im = np.zeros((5, 5))
vals = im_utils.neighbor_vals(im, 4, 2)
# make assertion
assert vals[0] == 0.
assert vals[1] == 0.
assert vals[3] == 0.
assert vals[5] == 0.
assert vals[6] == 0.
def simplify_skel(Iskel):
"""
This function iterates through all skeleton pixels pixels that have
connectivity > 2. It removes the pixel and checks if the
number of blobs has changed after removal. If so, the pixel is necessary to
maintain connectivity. Otherwise the pixel is not retained. It also adds
pixels to the centers of "+" cases, as this reduces the number
of branchpoints from 4 to 1.
Parameters
----------
Iskel : np.array
Image of the skeleton to simplify.
Returns
-------
Iskel : np.array
The simplified skeleton.
"""
Iskel = np.array(Iskel, dtype=np.uint8)
Ic = imu.im_connectivity(Iskel)
ypix, xpix = np.where(Ic > 2) # Get all pixels with connectivity > 2
for y, x in zip(ypix, xpix):
nv = imu.neighbor_vals(Iskel, x, y)
# Skip edge cases
if np.any(np.isnan(nv)) == True:
continue
# Create 3x3 array with center pixel removed
Inv = np.array(
[[nv[0], nv[1], nv[2]], [nv[3], 0, nv[4]], [nv[5], nv[6], nv[7]]],
dtype=bool)
# Check the connectivity after removing the pixel, set to zero if unchanged
Ilabeled = measure.label(Inv, background=0, connectivity=2)
if np.max(Ilabeled) == 1:
Iskel[y, x] = 0
# We simplify the network if we actually add pixels to the centers of
# "+" cases, so let's do that.
kern = np.array([[1, 10, 1], [10, 1, 10], [1, 10, 1]], dtype=np.uint8)
Iconv = cv2.filter2D(Iskel, -1, kern)
Iskel[Iconv == 40] = 1
return Iskel
def skel_simplify(Iskel):
"""
Another attempt to simplify skeletons. This method iterates through all
pixels that have connectivity > 2. It removes the pixel and checks if the
number of blobs has changed after removal. If so, the pixel is necessary to
maintain connectivity.
"""
Iskel = np.array(Iskel, dtype=np.uint8)
Ic = iu.im_connectivity(Iskel)
ypix, xpix = np.where(Ic > 2) # Get all pixels with connectivity > 2
for y, x in zip(ypix, xpix):
nv = iu.neighbor_vals(Iskel, x, y)
# Check for edge cases; skip them for now
if np.any(np.isnan(nv)) == True:
continue
# Create 3x3 array with center pixel removed
Inv = np.array(
[[nv[0], nv[1], nv[2]], [nv[3], 0, nv[4]], [nv[5], nv[6], nv[7]]],
dtype=np.bool)
# Check the connectivity after removing the pixel, set to zero if unchanged
Ilabeled = measure.label(Inv, background=0, connectivity=2)
if np.max(Ilabeled) == 1:
Iskel[y, x] = 0
# We simplify the network if we actually add pixels to the centers of
# "+" cases, so let's do that.
kern = np.array([[1, 10, 1], [10, 1, 10], [1, 10, 1]], dtype=np.uint8)
Iconv = cv2.filter2D(Iskel, -1, kern)
Iskel[Iconv == 40] = 1
return Iskel
