def test_wrong_input_type(self):
wrong_dtype = np.ones((3, 3))*255 #image with wrong dtype(int64)
not_gray_scale = np.ones((5, 5, 3), dtype = np.uint8) * 255 #image with 3 channels
with pytest.raises(TypeError, match='input image must be gray scale image as uint8 ndarray'):
crop_boundary_and_padding(wrong_dtype)
with pytest.raises(TypeError, match='input image must be gray scale image as uint8 ndarray'):
crop_boundary_and_padding(not_gray_scale)
def test_with_pad(self):
padded_in_0 = np.ones((3, 3), dtype = np.uint8)
res_0 = np.zeros((10,20), dtype = np.uint8)
res_0[2:5, 10:13] = padded_in_0
padded_in_1 = np.ones((3, 3), dtype = np.uint8)
res_1 = np.zeros((9,9), dtype = np.uint8)
res_1[3:6, 3:6] = padded_in_1
assert(np.array_equal(crop_boundary_and_padding(padded_in_0, [10,7,2,5]), res_0))
assert(np.array_equal(crop_boundary_and_padding(padded_in_1, 3), res_1))
def test_invalid_pad(self):
wrong_pad0 = [0,0,0,0,0]
wrong_pad1 = 15432514.315345
wrong_pad2 = [0]
in_im = np.zeros((100, 200), dtype = np.uint8)
in_im[30, 30] = 255
with pytest.raises(TypeError, match='padding in crop function must be int or list of size 4'):
crop_boundary_and_padding(in_im, padding = wrong_pad0)
with pytest.raises(TypeError, match='padding in crop function must be int or list of size 4'):
crop_boundary_and_padding(in_im, padding = wrong_pad1)
with pytest.raises(TypeError, match='padding in crop function must be int or list of size 4'):
crop_boundary_and_padding(in_im, padding = wrong_pad2)
def test_wrong_binarized_input(self):
in_im = np.ones((3,3), dtype=np.uint8)
big_mask = np.ones((4,4), dtype=np.uint8) * 255
small_mask = np.ones((2,2), dtype=np.uint8) * 255
wrong_type = np.ones((3,3)) * 255
with pytest.raises(ValueError, match='binarized image shape .* must meet in image shape .*'):
crop_boundary_and_padding(in_im, binarized=big_mask)
with pytest.raises(ValueError, match='binarized image shape .* must meet in image shape .*'):
crop_boundary_and_padding(in_im, binarized=small_mask)
with pytest.raises(TypeError, match='binarized mask must be uint8 ndarray'):
crop_boundary_and_padding(in_im, binarized=wrong_type)
def test_all_zero_im(self):
in_im = np.zeros((100, 200), dtype = np.uint8) #All zero input image with whatever shape
with pytest.raises(ValueError, match='In image for crop function is all zero'):
crop_boundary_and_padding(in_im)
def test_gray(self):
padded_in0, res_0 = self.gray_case0()
padded_in1, padded_binarized, res_1 = self.gray_case1()
assert (np.array_equal(crop_boundary_and_padding(padded_in0), res_0)) #without binarize input
assert (np.array_equal(crop_boundary_and_padding(padded_in1, binarized = padded_binarized), res_1)) #with binarize map
def test_binarized(self):
padded_in0, res_0 = self.binarized_case0()
padded_in1, res_1 = self.binarized_case1()
assert(np.array_equal(crop_boundary_and_padding(padded_in0), res_0))
assert(np.array_equal(crop_boundary_and_padding(padded_in1), res_1))
def test_out_type(self):
x = np.ones((3, 3), dtype=np.uint8)
out = crop_boundary_and_padding(x)
assert out.dtype == np.uint8
