def test_smaller_even_output(self):
from dfm.reframe import reframe
a = numpy.array([[1, 1, 1], [1, 2, 1], [1, 1, 1]])
expected = numpy.array([
[0, 0, 0, 0],
[0, 1, 1, 1],
[0, 1, 2, 1],
[0, 1, 1, 1],
])
out = reframe(a, width=4, height=4, x=2, y=2)
print('')
pprint(out)
assert_array_equal(out, expected)
def test_offcenter_larger(self):
from dfm.reframe import reframe
a = numpy.array([
[1, 1, 1, 1, 1],
[1, 2, 2, 2, 1],
[1, 2, 3, 2, 1],
[1, 2, 2, 2, 1],
[1, 1, 1, 1, 1],
])
expected = numpy.array([[2, 3, 2], [2, 2, 2], [1, 1, 1]])
out = reframe(a, width=3, height=3, x=1, y=0)
print('')
pprint(out)
assert_array_equal(out, expected)
def test_offcenter_smaller_cropped(self):
from dfm.reframe import reframe
a = numpy.array([[1, 1, 1], [1, 2, 1], [1, 1, 1]])
expected = numpy.array([
[0, 0, 0, 1, 1],
[0, 0, 0, 1, 2],
[0, 0, 0, 1, 1],
[0, 0, 0, 0, 0],
[0, 0, 0, 0, 0],
])
out = reframe(a, width=5, height=5, x=4, y=1)
print('')
pprint(out)
assert_array_equal(out, expected)
else:
raise ValueError('No convergence when finding local maxima')
features = pandas.DataFrame(features)
gabor_vects = numpy.full([features.shape[0], size**2], numpy.nan)
for feature in tqdm.tqdm(features.itertuples(), desc='construct features'):
f = feature.Index
kside = 1 + 2 * int(ceil(kernel_extent * feature.sigma))
kernel_params = dict(ksize=(kside, kside),
sigma=feature.sigma,
theta=feature.theta,
lambd=feature.lambd,
gamma=gamma,
ktype=cv2.CV_32F)
kernel = cv2.getGaborKernel(psi=0, **kernel_params)
gabor = reframe(kernel, width=size, height=size, x=feature.x, y=feature.y)
gabor_vects[f, :] = gabor.flatten()
print('determine explained variance by feature..')
covs = gabor_vects @ image.flatten()
areas = numpy.sum(numpy.abs(gabor_vects), axis=1)
expl_var = covs / areas
expl_var_ranks = numpy.flip(numpy.argsort(expl_var))
features['expl_var_rank'] = expl_var_ranks
## selection
selection = numpy.full_like(expl_var_ranks, False, dtype=bool)
img_vect = image.ravel()
selection[expl_var_ranks[0]] = True ## select first gabor
recon_vect = gabor_vects[selection, :].sum(axis=0)
recon_r = numpy.corrcoef(recon_vect, img_vect)[0, 1]