def test_load_ffi_files():
files = find_ffi_files(INPUT_DIR, camera=1)
img = load_ffi_fits(files[0])
assert (img.shape == (2048, 2048))
img, hdr = load_ffi_fits(files[0], return_header=True)
assert (img.shape == (2048, 2048))
def test_load_ffi_files(SHARED_INPUT_DIR):
files = u.find_ffi_files(SHARED_INPUT_DIR, camera=1)
img = u.load_ffi_fits(files[0])
assert (img.shape == (2048, 2048))
img, hdr = u.load_ffi_fits(files[0], return_header=True)
assert (img.shape == (2048, 2048))
def test_background(SHARED_INPUT_DIR):
"""Test of background estimator"""
# Load the first image in the input directory:
fname = find_ffi_files(SHARED_INPUT_DIR)[0]
img, hdr = load_ffi_fits(fname, return_header=True)
# Estimate the background:
bck, mask = fit_background(fname)
# Print some information:
print(fname)
print(bck.shape)
print(mask.shape)
# Check the sizes of the returned images:
assert (bck.shape == img.shape)
assert (mask.shape == img.shape)
def fit_background(image, catalog=None, flux_cutoff=8e4):
"""
Estimate background in Full Frame Image.
Parameters:
image (ndarray or string): Either the image as 2D ndarray or a path to FITS or NPY file where to load image from.
catalog (`astropy.table.Table` object): Catalog of stars in the image. Is not yet being used for anything.
flux_cutoff (float): Flux value at which any pixel above will be masked out of the background estimation.
Returns:
ndarray: Estimated background with the same size as the input image.
ndarray: Boolean array specifying which pixels was used to estimate the background (``True`` if pixel was used).
.. codeauthor:: Rasmus Handberg <*****@*****.**>
"""
# Load file:
if isinstance(image, np.ndarray):
img = image
elif isinstance(image, six.string_types):
if image.endswith('.npy'):
img = np.load(image)
else:
img = load_ffi_fits(image)
else:
raise ValueError(
"Input image must be either 2D ndarray or path to file.")
# Create mask
# TODO: Use the known locations of bright stars
mask = ~np.isfinite(img)
mask |= (img > flux_cutoff)
# Estimate the background:
sigma_clip = SigmaClip(sigma=3.0, iters=5)
bkg_estimator = SExtractorBackground()
bkg = Background2D(img, (64, 64),
filter_size=(3, 3),
sigma_clip=sigma_clip,
bkg_estimator=bkg_estimator,
mask=mask,
exclude_percentile=50)
return bkg.background, mask
def test_background(): """Test of background estimator""" # Load the first image in the input directory: INPUT_DIR = os.path.join(os.path.dirname(__file__), 'input', 'images') fname = find_ffi_files(INPUT_DIR)[0] img, hdr = load_ffi_fits(fname, return_header=True) # Estimate the background: bck, mask = fit_background(fname, ) # Print some information: print(fname) print(bck.shape) print(mask.shape) # Check the sizes of the returned images: assert(bck.shape == img.shape) assert(mask.shape == img.shape)
def test_imagemotion():
"""Test of ImageMovementKernel"""
# Load the first image in the input directory:
INPUT_DIR = os.path.join(os.path.dirname(__file__), 'input', 'images')
files = find_ffi_files(INPUT_DIR, camera=1, ccd=1)
fname = files[0]
# Load the image:
img = load_ffi_fits(fname)
# Create new image, moved down by one pixel:
#img2 = np.roll(img, 1, axis=0)
# Some positions across the image:
xx, yy = np.meshgrid(
np.linspace(0, img.shape[0], 5, dtype='float64'),
np.linspace(0, img.shape[1], 5, dtype='float64'),
)
xy = np.column_stack((xx.flatten(), yy.flatten()))
print("Positions to be tested:")
print(xy)
desired1 = np.zeros_like(xy)
desired2 = np.zeros_like(xy)
desired2[:, 1] = 1
for warpmode in ('unchanged', 'translation', 'euclidian'):
print("Testing warpmode=" + warpmode)
# Create ImageMovementKernel instance:
imk = ImageMovementKernel(image_ref=img, warpmode=warpmode)
# Calculate kernel for the same image:
kernel = imk.calc_kernel(img)
print("Kernel:")
print(kernel)
assert (len(kernel) == imk.n_params)
# Calculate the new positions based on the kernel:
delta_pos = imk(xy, kernel)
print("Extracted movements:")
print(delta_pos)
assert (delta_pos.shape == xy.shape)
# The movements should all be very close to zero,
# since we used the same image as the reference:
np.testing.assert_allclose(delta_pos, desired1, atol=1e-5, rtol=1e-5)
"""
kernel = imk.calc_kernel(img2)
print("Kernel 2:")
print(kernel)
# Calculate the new positions based on the kernel:
delta_pos = imk(xy, kernel)
print("Extracted movements:")
print(delta_pos)
assert(delta_pos.shape == xy.shape)
# The movements should all be very close to zero,
# since we used the same image as the reference:
# FIXME: Would LOVE this to be more accurate!
np.testing.assert_allclose(delta_pos, desired2, atol=1e-3, rtol=1e-2)
"""
print("Done")
def test_imagemotion(SHARED_INPUT_DIR, warpmode):
"""Test of ImageMovementKernel"""
print("Testing warpmode=" + warpmode)
# Load the first image in the input directory:
files = find_ffi_files(os.path.join(SHARED_INPUT_DIR, 'images'),
camera=1,
ccd=1)
fname = files[0]
# Load the image:
img = load_ffi_fits(fname)
# Create new image, moved down by one pixel:
#img2 = np.roll(img, 1, axis=0)
# Some positions across the image:
xx, yy = np.meshgrid(
np.linspace(0, img.shape[0], 5, dtype='float64'),
np.linspace(0, img.shape[1], 5, dtype='float64'),
)
xy = np.column_stack((xx.flatten(), yy.flatten()))
print("Positions to be tested:")
print(xy)
desired1 = np.zeros_like(xy)
desired2 = np.zeros_like(xy)
desired2[:, 1] = 1
# Create ImageMovementKernel instance:
imk = ImageMovementKernel(image_ref=img, warpmode=warpmode)
# If calling interpolate before defining kernels, we should get an error:
with pytest.raises(ValueError):
imk.interpolate(1234, 1234)
# Calculate kernel for the same image:
kernel = imk.calc_kernel(img)
print("Kernel:")
print(kernel)
assert (len(kernel) == imk.n_params)
# Calculate the new positions based on the kernel:
delta_pos = imk(xy, kernel)
print("Extracted movements:")
print(delta_pos)
assert (delta_pos.shape == xy.shape)
# The movements should all be very close to zero,
# since we used the same image as the reference:
np.testing.assert_allclose(delta_pos, desired1, atol=1e-5, rtol=1e-5)
#
with pytest.raises(ValueError) as excinfo:
imk.load_series([1234, 1235], kernel)
print(excinfo.value)
assert str(excinfo.value).startswith('Wrong shape of kernels.')
"""
kernel = imk.calc_kernel(img2)
print("Kernel 2:")
print(kernel)
# Calculate the new positions based on the kernel:
delta_pos = imk(xy, kernel)
print("Extracted movements:")
print(delta_pos)
assert(delta_pos.shape == xy.shape)
# The movements should all be very close to zero,
# since we used the same image as the reference:
# FIXME: Would LOVE this to be more accurate!
np.testing.assert_allclose(delta_pos, desired2, atol=1e-3, rtol=1e-2)
"""
print("Done")