def test_correlate_sparse_invalid_kernel(mode):
image = np.array([[0, 0, 1, 3, 5], [0, 1, 4, 3, 4], [1, 2, 5, 4, 1],
[2, 4, 5, 2, 1], [4, 5, 1, 0, 0]],
dtype=np.float)
# invalid kernel size
invalid_kernel = np.array([0, 1, 2, 4]).reshape((2, 2))
with testing.raises(ValueError):
correlate_sparse(image, invalid_kernel, mode=mode)
def test_correlate_sparse(mode):
image = np.array([[0, 0, 1, 3, 5], [0, 1, 4, 3, 4], [1, 2, 5, 4, 1],
[2, 4, 5, 2, 1], [4, 5, 1, 0, 0]],
dtype=np.float)
kernel = np.array([0, 1, 2, 4, 8, 16, 32, 64, 128]).reshape((3, 3))
cs_output = correlate_sparse(image, kernel, mode=mode)
ndi_output = ndi.correlate(image, kernel, mode=mode)
assert_equal(cs_output, ndi_output)
def test_correlate_sparse_valid_mode():
image = np.array([[0, 0, 1, 3, 5], [0, 1, 4, 3, 4], [1, 2, 5, 4, 1],
[2, 4, 5, 2, 1], [4, 5, 1, 0, 0]],
dtype=float)
kernel = np.array([0, 1, 2, 4, 8, 16, 32, 64, 128]).reshape((3, 3))
cs_output = correlate_sparse(image, kernel, mode="valid")
ndi_output = ndi.correlate(image, kernel, mode='wrap')
ndi_output = ndi_output[1:4, 1:4]
assert_equal(cs_output, ndi_output)
def test_correlate_sparse(mode, dtype):
image = np.array([[0, 0, 1, 3, 5],
[0, 1, 4, 3, 4],
[1, 2, 5, 4, 1],
[2, 4, 5, 2, 1],
[4, 5, 1, 0, 0]], dtype=dtype)
kernel = np.array([0, 1, 2, 4, 8, 16, 32, 64, 128]).reshape((3, 3))
cs_output = correlate_sparse(image, kernel, mode=mode)
assert cs_output.dtype == _supported_float_type(image.dtype)
ndi_output = ndi.correlate(
image.astype(float, copy=False), kernel, mode=mode
)
assert_equal(cs_output, ndi_output)
