def test_downsampling_beyond_dimensions_produces_valueerror(scale):
image = ImageData(channels=np.arange(24).reshape(1, 4, 6), pixel_resolution_um=12)
if scale <= 0:
with pytest.raises(ValueError, match=r".* positive.*"):
image.resample(scale)
elif scale < 0.25:
with pytest.raises(ValueError, match=r".* small.*"):
image.resample(scale)
def test_downsampling_image_produces_correct_resolution_and_data_shape(scale):
image = ImageData(channels=np.arange(24).reshape(1, 6, 4), pixel_resolution_um=12)
image2 = image.resample(scale)
assert image2.pixel_resolution_um == (image.pixel_resolution_um / scale)
assert image2.num_channels == image.num_channels
assert approx(image.width * scale == image2.width, abs=1)
assert approx(image.height * scale == image2.height, abs=1)
def test_downsampling_image_produces_correct_resolution_and_data_shape(
to_resolution):
from_resolution = 12
image = ImageData(channels=np.arange(24).reshape(1, 6, 4),
pixel_resolution_um=from_resolution)
image2 = image.resample(resolution_um=to_resolution)
assert image2.pixel_resolution_um == to_resolution
assert image2.num_channels == image.num_channels
scale_ratio = from_resolution / to_resolution
assert approx(image.width * scale_ratio == image2.width, abs=1)
assert approx(image.height * scale_ratio == image2.height, abs=1)
def test_downsampling_image_produces_im_with_similar_statistical_properties_as_original(scale):
image = ImageData(channels=np.random.random((2, 120, 120)), pixel_resolution_um=12)
im2 = image.resample(scale)
assert np.all(np.isclose(image.channels.mean(), im2.channels.mean(), rtol=3e-2))
assert np.all(np.isclose(image.channels.std(), im2.channels.std(), rtol=3e-2))
def test_downsampling_beyond_dimensions_produces_valueerror(to_resolution):
image = ImageData(channels=np.arange(24).reshape(1, 4, 6),
pixel_resolution_um=12)
with pytest.raises(ValueError, match=r".* positive.*"):
image.resample(resolution_um=to_resolution)