# Copyright (C) 2022 ISIS Rutherford Appleton Laboratory UKRI
# SPDX - License - Identifier: GPL-3.0-or-later
import math
import numpy as np
import pytest
from numpy import testing as npt, copy
import mantidimaging.test_helpers.unit_test_helper as th
from mantidimaging.core.operations.rescale import RescaleFilter
from mantidimaging.test_helpers.qt_mocks import MockQSpinBox, MockQComboBox
[docs]
@pytest.mark.parametrize('value', [255.0, 65535.0, 2147483647.0])
def test_rescale(value):
images = th.generate_images((10, 100, 100))
images.data[0:3] = -100
images.data[3:6] = 0.5
images.data[6:10] = 1.0
expected_min_input = 0.0
images = RescaleFilter.filter_func(images,
min_input=expected_min_input,
max_input=images.data.max(),
max_output=value)
# below min_input has been clipped to 0
npt.assert_equal(0, images.data[0:3])
npt.assert_equal(images.data[3:6], value / 2)
npt.assert_equal(images.data[6:10], value)
[docs]
def test_execute_wrapper_no_preset():
min_input_value = 12
max_input_value = 34
min_input = MockQSpinBox(min_input_value)
max_input = MockQSpinBox(max_input_value)
max_output = MockQSpinBox(420.0)
preset = MockQComboBox('None')
partial = RescaleFilter.execute_wrapper(min_input, max_input, max_output, preset)
assert partial.keywords['min_input'] == min_input_value
assert partial.keywords['max_input'] == max_input_value
assert partial.keywords['max_output'] == 420.0
[docs]
@pytest.mark.parametrize('type, expected_max', [
('int8', 255.0),
('int16', 65535),
('int32', 2147483647.0),
])
def test_execute_wrapper_with_preset(type: str, expected_max: float):
min_input_value = 12
max_input_value = 34
min_input = MockQSpinBox(min_input_value)
max_input = MockQSpinBox(max_input_value)
max_output = MockQSpinBox(420.0) # this value is overridden by preset
preset = MockQComboBox(type)
partial = RescaleFilter.execute_wrapper(min_input, max_input, max_output, preset) # type: ignore
assert partial.keywords['min_input'] == min_input_value
assert partial.keywords['max_input'] == max_input_value
assert partial.keywords['max_output'] == expected_max
[docs]
def test_scale_single_image():
images = th.generate_images((2, 100, 100))
images.data[0:2] = np.arange(-1, 1, step=0.0002).reshape(100, 100)
scaled_image = RescaleFilter.filter_array(copy(images.data[0]),
min_input=images.data[0].min(),
max_input=images.data[0].max(),
max_output=65535)
assert scaled_image.min() == 0
assert scaled_image.max() == 65535
[docs]
def test_scale_single_image_bad_offset():
images = th.generate_images((2, 100, 100))
try:
RescaleFilter.filter_array(copy(images.data[0]), min_input=-5000, max_input=5000, max_output=65535)
except ValueError:
pass
except Exception as e:
assert False, f"Unexpected exception was triggered: {e}"
[docs]
@pytest.mark.parametrize('value', [255.0, 65535.0, 2147483647.0])
def test_rescale_ignores_nans(value):
images = th.generate_images((10, 100, 100))
images.data[0:3] = -100.0
images.data[3:5] = 0.5
images.data[6][0:10] = np.nan
images.data[7:10] = 1.0
expected_min_input = 0.0
images = RescaleFilter.filter_func(images,
min_input=expected_min_input,
max_input=np.nanmax(images.data),
max_output=value)
# below min_input has been clipped to 0
npt.assert_equal(0, images.data[0:3])
npt.assert_equal(images.data[3:5], value / 2)
npt.assert_equal(images.data[7:10], value)
assert all([math.isnan(x) for x in images.data[6][0:10].flatten()])
if __name__ == "__main__":
import pytest
pytest.main([__file__])