def test_threshold_local_gaussian(self, ndim):
ref = np.array([[False, False, False, False, True],
[False, False, True, False, True],
[False, False, True, True, False],
[False, True, True, False, False],
[True, True, False, False, False]])
if ndim == 2:
image = self.image
block_sizes = [3, (3, ) * image.ndim]
else:
image = np.stack((self.image, ) * 5, axis=-1)
ref = np.stack((ref, ) * 5, axis=-1)
block_sizes = [
3, (3, ) * image.ndim, (3, ) * (image.ndim - 1) + (1, )
]
for block_size in block_sizes:
out = threshold_local(image,
block_size,
method='gaussian',
mode='reflect')
assert_equal(ref, image > out)
out = threshold_local(image,
3,
method='gaussian',
mode='reflect',
param=1 / 3)
assert_equal(ref, image > out)
def test_threshold_local_gaussian(self):
ref = np.array([[False, False, False, False, True],
[False, False, True, False, True],
[False, False, True, True, False],
[False, True, True, False, False],
[True, True, False, False, False]])
out = threshold_local(self.image, 3, method='gaussian')
assert_equal(ref, self.image > out)
out = threshold_local(self.image, 3, method='gaussian', param=1. / 3.)
assert_equal(ref, self.image > out)
def test_threshold_local_gaussian(self):
ref = np.array(
[[False, False, False, False, True],
[False, False, True, False, True],
[False, False, True, True, False],
[False, True, True, False, False],
[ True, True, False, False, False]]
)
out = threshold_local(self.image, 3, method='gaussian')
assert_equal(ref, self.image > out)
out = threshold_local(self.image, 3, method='gaussian',
param=1./3.)
assert_equal(ref, self.image > out)
def test_threshold_local_median(self):
ref = np.array([[False, False, False, False, True],
[False, False, True, False, False],
[False, False, True, False, False],
[False, False, True, True, False],
[False, True, False, False, False]])
out = threshold_local(self.image, 3, method='median')
assert_equal(ref, self.image > out)
def test_threshold_local_median_constant_mode(self):
out = threshold_local(self.image, 3, method='median',
mode='constant', cval=20)
expected = np.array([[20., 1., 3., 4., 20.],
[ 1., 1., 3., 4., 4.],
[ 2., 2., 4., 4., 4.],
[ 4., 4., 4., 1., 2.],
[20., 5., 5., 2., 20.]])
assert_equal(expected, out)
def test_threshold_local_median_constant_mode(self):
out = threshold_local(self.image, 3, method='median',
mode='constant', cval=20)
expected = np.array([[20., 1., 3., 4., 20.],
[ 1., 1., 3., 4., 4.],
[ 2., 2., 4., 4., 4.],
[ 4., 4., 4., 1., 2.],
[20., 5., 5., 2., 20.]])
assert_equal(expected, out)
def test_threshold_local_equals_adaptive(self):
def func(arr):
return arr.sum() / arr.shape[0]
with expected_warnings(['deprecated', 'return value']):
thresholded_original = threshold_adaptive(self.image, 3,
method='generic',
param=func)
threshold_new = threshold_local(self.image, 3, method='generic',
param=func)
assert_equal(thresholded_original, self.image > threshold_new)
def test_threshold_local_equals_adaptive(self):
def func(arr):
return arr.sum() / arr.shape[0]
with expected_warnings(['deprecated', 'return value']):
thresholded_original = threshold_adaptive(self.image, 3,
method='generic',
param=func)
threshold_new = threshold_local(self.image, 3, method='generic',
param=func)
assert_equal(thresholded_original, self.image > threshold_new)
def test_threshold_local_median(self):
ref = np.array(
[[False, False, False, False, True],
[False, False, True, False, False],
[False, False, True, False, False],
[False, False, True, True, False],
[False, True, False, False, False]]
)
out = threshold_local(self.image, 3, method='median')
assert_equal(ref, self.image > out)
def test_threshold_local_median(self, ndim):
ref = np.array([[False, False, False, False, True],
[False, False, True, False, False],
[False, False, True, False, False],
[False, False, True, True, False],
[False, True, False, False, False]])
if ndim == 2:
image = self.image
else:
image = np.stack((self.image, ) * 5, axis=-1)
ref = np.stack((ref, ) * 5, axis=-1)
out = threshold_local(image, 3, method='median', mode='reflect')
assert_equal(ref, image > out)
def test_threshold_local_mean(self, ndim):
ref = np.array(
[[False, False, False, False, True],
[False, False, True, False, True],
[False, False, True, True, False],
[False, True, True, False, False],
[ True, True, False, False, False]]
)
if ndim == 2:
image = self.image
block_sizes = [3, (3,) * image.ndim]
else:
image = np.stack((self.image, ) * 5, axis=-1)
ref = np.stack((ref, ) * 5, axis=-1)
# Given the same data at each z location, the following block sizes
# will all give an equivalent result.
block_sizes = [3, (3,) * image.ndim,
(3,) * (image.ndim - 1) + (1,)]
for block_size in block_sizes:
out = threshold_local(image, block_size, method='mean',
mode='reflect')
assert_equal(ref, image > out)
def test_adaptive_even_block_size_error():
img = data.camera()
with pytest.raises(ValueError):
threshold_local(img, block_size=4)
def test_adaptive_even_block_size_error():
img = data.camera()
with pytest.raises(ValueError):
threshold_local(img, block_size=4)
from skimage.color import rgb2gray
import os
#image = data.camera()
filename = "fake_images/Screen Shot 2019-08-22 at 11.06.38 AM.png"
from skimage import io
moon = io.imread(filename)
image = rgb2gray(moon)
global_thresh = threshold_otsu(image)
binary_global = image > global_thresh
block_size = 13
binary_adaptive = threshold_local(image, block_size, offset=10)
fig, axes = plt.subplots(nrows=3, figsize=(7, 8))
ax0, ax1, ax2 = axes
plt.gray()
ax0.imshow(image)
ax0.set_title('Image')
ax1.imshow(binary_global)
ax1.set_title('Global thresholding')
ax2.imshow(binary_adaptive)
ax2.set_title('Adaptive thresholding')
for ax in axes:
def test_threshold_local_invalid_block_size(self, block_size):
# len(block_size) != image.ndim
with pytest.raises(ValueError):
threshold_local(self.image, block_size, method='mean')
def test_local_even_block_size_error():
img = data.camera()
with testing.raises(ValueError):
threshold_local(img, block_size=4)
def test_local_even_block_size_error():
img = data.camera()
with testing.raises(ValueError):
threshold_local(img, block_size=4)
