def test_Filament2D_iteratbranchprune_leavenone():
'''
Add multiple branches that should be pruned, leaving a single branch
skeleton.
Longest path will have a length of 7 + sqrt(2)
'''
pixels = (np.array([0, 2, 1, 4, 1, 2, 3, 1, 4, 5, 1, 1, 1]) + 1,
np.array([0, 0, 1, 1, 2, 2, 2, 3, 3, 3, 4, 5, 6]) + 1)
prune_pixels = (np.array([1, 1, 1, 1, 1, 2, 3, 4, 5]) + 1,
np.array([2, 3, 4, 5, 6, 2, 2, 3, 3]) + 1)
shape = (7, 8)
image = np.zeros(shape)
image[pixels] = 2.
fil = Filament2D(pixels)
assert all([(out == inp).all()
for out, inp in zip(fil.pixel_coords, pixels)])
assert fil.pixel_extents == [(1, 1), (pixels[0].max(), pixels[1].max())]
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels] = True
# Now run the skeleton analysis
fil.skeleton_analysis(image,
prune_criteria='length',
branch_thresh=2 * u.pix)
for pix_exp, pix_have in zip(prune_pixels, fil.pixel_coords):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
# The extra branch should now be removed.
mask_prune = fil.skeleton(pad_size=1)
pix = np.where(mask_prune)
for pix_exp, pix_have in zip(prune_pixels, pix):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
assert (fil.skeleton() == fil.skeleton(out_type='longpath')).all()
# The pruned graph should be reduced to a single node representing the
# remaining branch
assert list(fil.graph.nodes()) == [1]
assert dict(fil.graph.degree()) == {1: 0}
# The branch properties should match those of the longest path
assert fil.branch_properties['length'] == fil.length()
assert fil.branch_properties['number'] == 1
assert fil.branch_properties['intensity'] == fil.median_brightness(image)
for pix_exp, pix_have in zip(prune_pixels,
fil.branch_properties['pixels'][0].T):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
def test_Filament2D_with_distance():
pixels = (np.array([0, 0, 0]), np.array([0, 1, 2]))
image = np.zeros((1, 3))
image[0, :] = 2.
mywcs = WCS()
mywcs.wcs.ctype = ['GLON-CAR', 'GLAT-CAR']
mywcs.wcs.cunit = ['deg', 'deg']
fil = Filament2D(pixels, wcs=mywcs, distance=100 * u.pc)
fil.skeleton_analysis(image)
# Check the length
assert fil.length().value == 2.0
assert fil.length().unit == u.pix
assert fil.length(unit=u.deg) == 2.0 * u.deg
assert fil.length(unit=u.pc) == (2.0 * u.deg).to(u.rad).value * 100 * u.pc
def test_Filament2D_onebranchprune_wpadding():
'''
Longest path will be in a straight line. Expect length of 8
'''
from fil_finder.filament import Filament2D
pixels = (np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2]) + 1,
np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 3, 3]) + 1)
prune_pixels = (np.array([0, 0, 0, 0, 0, 0, 0, 0, 0]) + 1,
np.array([0, 1, 2, 3, 4, 5, 6, 7, 8]) + 1)
shape = (5, 11)
image = np.zeros(shape)
image[pixels] = 2.
fil = Filament2D(pixels)
assert all([(out == inp).all()
for out, inp in zip(fil.pixel_coords, pixels)])
assert fil.pixel_extents == [(1, 1), (pixels[0].max(), pixels[1].max())]
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels] = True
mask = fil.skeleton(pad_size=1)
for pix_exp, pix_have in zip(pixels, fil.pixel_coords):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
pix = np.where(mask)
for pix_exp, pix_have in zip(pixels, pix):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
# Now run the skeleton analysis
fil.skeleton_analysis(image,
prune_criteria='length',
branch_thresh=2 * u.pix)
for pix_exp, pix_have in zip(prune_pixels, fil.pixel_coords):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
# The extra branch should now be removed.
mask_prune = fil.skeleton(pad_size=1)
pix = np.where(mask_prune)
for pix_exp, pix_have in zip(prune_pixels, pix):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
assert (fil.skeleton() == fil.skeleton(out_type='longpath')).all()
# The pruned graph should be reduced to a single node representing the
# remaining branch
assert list(fil.graph.nodes()) == [1]
assert dict(fil.graph.degree()) == {1: 0}
# The branch properties should match those of the longest path
assert fil.branch_properties['length'] == fil.length()
assert fil.branch_properties['number'] == 1
assert fil.branch_properties['intensity'] == fil.median_brightness(image)
for pix_exp, pix_have in zip(prune_pixels,
fil.branch_properties['pixels'][0].T):
assert pix_exp.size == pix_have.size
assert (pix_exp == pix_have).all()
def test_Filament2D_onebranch_wpadding():
'''
Longest path will be in a straight line. Expect length of 8
'''
pixels = (np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2]) + 1,
np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 3, 3]) + 1)
shape = (5, 11)
image = np.zeros(shape)
image[pixels] = 2.
fil = Filament2D(pixels)
assert all([(out == inp).all()
for out, inp in zip(fil.pixel_coords, pixels)])
assert fil.pixel_extents == [(1, 1), (pixels[0].max(), pixels[1].max())]
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels] = True
mask = fil.skeleton(pad_size=1)
assert (mask == mask_expect).all()
# Long path mask should fail without skeleton_analysis
with pytest.raises(AttributeError):
fil.skeleton(out_type='longpath')
# Now run the skeleton analysis
fil.skeleton_analysis(image)
# Remove the one branch pixel
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels[0][:-2], pixels[1][:-2]] = True
pad = 1
mask = fil.skeleton(out_type='longpath', pad_size=pad)
mask_expect_slice = fil.image_slicer(mask_expect, mask.shape, pad_size=pad)
assert (mask == mask_expect_slice).all()
# Check the length
assert fil.length().value == 8.0
assert fil.length().unit == u.pix
# Angular and physical conversion should fail b/c no WCS or distance is
# given
with pytest.raises(AttributeError):
fil.length(unit=u.deg)
with pytest.raises(AttributeError):
fil.length(unit=u.pc)
# The branches should all have the same average intensity
for branch_int in fil.branch_properties['intensity']:
assert branch_int == 2.0
# Finally, make sure the transpose shape gives the same length, masks
fil = Filament2D((pixels[1], pixels[0]))
assert all([(out == inp).all()
for out, inp in zip(fil.pixel_coords, pixels[::-1])])
assert fil.pixel_extents == [(1, 1), (pixels[1].max(), pixels[0].max())]
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels] = True
mask = fil.skeleton(pad_size=1)
assert (mask == mask_expect.T).all()
fil.skeleton_analysis(image.T)
assert fil.length().value == 8.0
mask_expect = np.zeros(shape, dtype=bool)
mask_expect[pixels[0][:-2], pixels[1][:-2]] = True
mask = fil.skeleton(out_type='longpath', pad_size=pad)
assert (mask == mask_expect.T).all()
def test_Filament2D():
pixels = (np.array([0, 0, 0]), np.array([0, 1, 2]))
image = np.zeros((1, 3))
image[0, :] = 2.
fil = Filament2D(pixels)
assert all([(out == inp).all()
for out, inp in zip(fil.pixel_coords, pixels)])
assert fil.pixel_extents == [(0, 0), (0, 2)]
assert fil.position() == [0 * u.pix, 1 * u.pix]
# Should return pixels again because no WCS info is given
with warnings.catch_warnings(record=True) as w:
assert fil.position(world_coord=True) == [0 * u.pix, 1 * u.pix]
assert len(w) == 1
assert w[0].category == UserWarning
assert str(w[0].message) == ("No WCS information given. Returning pixel"
" position.")
mask_expect = np.zeros((1, 3), dtype=bool)
mask_expect[pixels] = True
mask = fil.skeleton()
assert (mask == mask_expect).all()
# Test out the padding
pad = 1
mask_expect = np.zeros((1 + 2 * pad, 3 + 2 * pad), dtype=bool)
mask_expect[pixels[0] + pad, pixels[1] + pad] = True
mask = fil.skeleton(pad_size=pad)
assert (mask == mask_expect).all()
# Long path mask should fail without skeleton_analysis
with pytest.raises(AttributeError):
fil.skeleton(out_type='longpath')
# Now run the skeleton analysis
fil.skeleton_analysis(image)
mask = fil.skeleton(out_type='longpath', pad_size=pad)
assert (mask == mask_expect).all()
# Check the length
assert fil.length().value == 2.0
assert fil.length().unit == u.pix
# Check the intersection and end points
assert len(fil.intersec_pts) == 0
assert fil.intersec_pts == []
assert len(fil.end_pts) == 2
assert fil.end_pts[0] == (0, 0)
assert fil.end_pts[1] == (0, 2)
# Angular and physical conversion should fail b/c no WCS or distance is
# given
with pytest.raises(AttributeError):
fil.length(unit=u.deg)
with pytest.raises(AttributeError):
fil.length(unit=u.pc)
# Test pickling
fil.to_pickle("pickled_fil.pkl")
loaded_fil = Filament2D.from_pickle("pickled_fil.pkl")
# Compare a few properties
assert (loaded_fil.length() == fil.length()).all()
assert (loaded_fil.skeleton(out_type='longpath',
pad_size=pad) == mask_expect).all()
import os
os.remove('pickled_fil.pkl')