def test_multiscale_basic_features_channel():
img = np.zeros((10, 10, 5))
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
n_sigmas = 2
features = multiscale_basic_features(img,
sigma_min=1,
sigma_max=2,
multichannel=True)
assert features.shape[-1] == 5 * n_sigmas * 4
assert features.shape[:-1] == img.shape[:-1]
# Consider last axis as spatial dimension
features = multiscale_basic_features(img, sigma_min=1, sigma_max=2)
assert features.shape[-1] == n_sigmas * 5
assert features.shape[:-1] == img.shape
def test_multiscale_basic_features_gray(edges, texture):
img = np.zeros((20, 20))
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
features = multiscale_basic_features(img, edges=edges, texture=texture)
n_sigmas = 6
intensity = True
assert features.shape[-1] == (
n_sigmas * (int(intensity) + int(edges) + 2 * int(texture))
)
assert features.shape[:-1] == img.shape[:]
def test_multiscale_basic_features_deprecated_multichannel():
img = np.zeros((10, 10, 5))
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
n_sigmas = 2
with expected_warnings(["`multichannel` is a deprecated argument"]):
features = multiscale_basic_features(img, sigma_min=1, sigma_max=2,
multichannel=True)
assert features.shape[-1] == 5 * n_sigmas * 4
assert features.shape[:-1] == img.shape[:-1]
# repeat prior test, but check for positional multichannel warning
with expected_warnings(["Providing the `multichannel` argument"]):
multiscale_basic_features(img, True, sigma_min=1, sigma_max=2)
assert features.shape[-1] == 5 * n_sigmas * 4
assert features.shape[:-1] == img.shape[:-1]
# Consider last axis as spatial dimension
features = multiscale_basic_features(img, sigma_min=1, sigma_max=2)
assert features.shape[-1] == n_sigmas * 5
assert features.shape[:-1] == img.shape
def test_multiscale_basic_features(edges, texture):
img = np.zeros((20, 20, 3))
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
with expected_warnings(["`multichannel` is a deprecated argument"]):
features = multiscale_basic_features(img, edges=edges, texture=texture,
multichannel=True)
n_sigmas = 6
intensity = True
assert features.shape[-1] == 3 * n_sigmas * (int(intensity) + int(edges) + 2 * int(texture))
assert features.shape[:-1] == img.shape[:-1]
def test_multiscale_basic_features_channel_axis(channel_axis):
num_channels = 5
shape_spatial = (10, 10)
ndim = len(shape_spatial)
shape = tuple(
np.insert(shape_spatial, channel_axis % (ndim + 1), num_channels)
)
img = np.zeros(shape)
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
n_sigmas = 2
# features for all channels are concatenated along the last axis
features = multiscale_basic_features(img, sigma_min=1, sigma_max=2,
channel_axis=channel_axis)
assert features.shape[-1] == 5 * n_sigmas * 4
assert features.shape[:-1] == np.moveaxis(img, channel_axis, -1).shape[:-1]
# Consider channel_axis as spatial dimension
features = multiscale_basic_features(img, sigma_min=1, sigma_max=2)
assert features.shape[-1] == n_sigmas * 5
assert features.shape[:-1] == img.shape
def test_trainable_segmentation_multichannel():
img = np.zeros((20, 20, 3))
img[:10] = 1
img += 0.05 * np.random.randn(*img.shape)
labels = np.zeros_like(img[..., 0], dtype=np.uint8)
labels[:2] = 1
labels[-2:] = 2
clf = DummyNNClassifier()
features = multiscale_basic_features(
img,
edges=False,
texture=False,
sigma_min=0.5,
sigma_max=2,
channel_axis=-1,
)
clf = fit_segmenter(labels, features, clf)
out = predict_segmenter(features, clf)
assert np.all(out[:10] == 1)
assert np.all(out[10:] == 2)
