def test_q2b_apply_lut_to_valid_colour_image():
# Same logic as previous test, just with a colour image.
plane = np.reshape(np.arange(256, dtype=np.uint8), (16, 16))
img = np.dstack((plane, plane, plane))
lut = np.arange(256, dtype=np.uint8) // 2
plane = np.reshape(lut, (16, 16))
expected = np.dstack((plane, plane, plane))
out = point_operators.apply_lut(img, lut)
assert_array_equal(out, expected)
def test_q2a_apply_lut_to_valid_greyscale_image():
# Use an image that contains all of the 8-bit values and use a LUT to
# perform a simple 1/2 scaling. The combination of the image's contents and
# the LUT's size means that the output image will be identical to the LUT
# after reshaping.
img = np.reshape(np.arange(256, dtype=np.uint8), (16, 16))
lut = np.arange(256, dtype=np.uint8) // 2
expected = np.reshape(lut, (16, 16))
out = point_operators.apply_lut(img, lut)
assert_array_equal(out, expected)
def test_q3c_exposure_adjustment(tmp_path):
img = skimage.data.coffee()
lut = point_operators.adjust_exposure(2.2)
out = point_operators.apply_lut(img, lut)
_plot_transfer_function(lut, tmp_path)
imsave(tmp_path / 'original.png', img)
imsave(tmp_path / 'processed.png', out)
# Compare against a reference image.
ref = imread(pathlib.Path() / 'samples' / 'reference' / 'exposure.png')
assert_array_equal(out, ref)
def test_q3d_log_transform(tmp_path):
img = skimage.data.hubble_deep_field()
img = img_as_ubyte(rgb2gray(img))
lut = point_operators.log_transform()
out = point_operators.apply_lut(img, lut)
_plot_transfer_function(lut, tmp_path)
imsave(tmp_path / 'original.png', img)
imsave(tmp_path / 'processed.png', out)
# Compare against a reference image.
ref = imread(pathlib.Path() / 'samples' / 'reference' /
'log-transform.png')
assert_array_equal(out, ref)
def _compare_stats_after_applying_lut(img, lut, folder):
'''Compare statistics after applying the LUT to an image.
Both the original and processed image will be saved to the specified folder.
Parameters
----------
img : numpy.ndarray
input image
lut : numpy.ndarray
look-up table
folder : path-like object
path to folder where the original image and adjusted image can be saved
Returns
-------
original_stats : ``(brightness, contrast)``
original image statistics
processed_stats : ``(brightness, contrast)``
post-application image statistics
'''
img = rgb2gray(img)
img = img_as_ubyte(img)
original_stats = _Stats()
original_stats.brightness = analysis.estimate_brightness(img)
original_stats.contrast = analysis.estimate_contrast(img)
processed = point_operators.apply_lut(img, lut)
processed_stats = _Stats()
processed_stats.brightness = analysis.estimate_brightness(processed)
processed_stats.contrast = analysis.estimate_contrast(processed)
imsave(folder / 'original.png', img)
imsave(folder / 'processed.png', processed)
return original_stats, processed_stats
def test_q2e_invalid_lut_length_raises_error():
lut = np.arange(10, dtype=np.uint8)
img = np.ones((10, 10), dtype=np.uint8)
with pytest.raises(ValueError):
point_operators.apply_lut(img, lut)
def test_q2d_invalid_lut_type_raises_error():
lut = np.arange(256)
img = np.ones((10, 10), dtype=np.uint8)
with pytest.raises(TypeError):
point_operators.apply_lut(img, lut)