def filter_window_polar(img, wsize, fun, rscale, random=False):
r"""Apply a filter of an approximated square window of half size `fsize`
on a given polar image `img`.
Parameters
----------
img : :class:`numpy:numpy.ndarray`
2d array of values to which the filter is to be applied
wsize : float
Half size of the window centred on the pixel [m]
fun : string
name of the 1d filter from :mod:`scipy:scipy.ndimage`
rscale : float
range [m] scale of the polar grid
random: bool
True to use random azimuthal size to avoid long-term biases.
Returns
-------
output : :class:`numpy:numpy.ndarray`
Array with the same shape as `img`, containing the filter's results.
"""
ascale = 2 * np.pi / img.shape[0]
data_filtered = np.empty(img.shape, dtype=img.dtype)
fun = getattr(filters, "%s_filter1d" % fun)
nbins = img.shape[-1]
ranges = np.arange(nbins) * rscale + rscale / 2
asize = ranges * ascale
if random:
na = prob_round(wsize / asize).astype(int)
else:
na = np.fix(wsize / asize + 0.5).astype(int)
# Maximum of adjacent azimuths (higher close to the origin) to
# increase performance
na[na > 20] = 20
sr = np.fix(wsize / rscale + 0.5).astype(int)
for sa in np.unique(na):
imax = np.where(na >= sa)[0][-1] + 1
imin = np.where(na <= sa)[0][0]
if sa == 0:
data_filtered[:, imin:imax] = img[:, imin:imax]
imin2 = max(imin - sr, 0)
imax2 = min(imax + sr, nbins)
temp = img[:, imin2:imax2]
temp = fun(temp, size=2 * sa + 1, mode='wrap', axis=0)
temp = fun(temp, size=2 * sr + 1, axis=1)
imin3 = imin - imin2
imax3 = imin3 + imax - imin
data_filtered[:, imin:imax] = temp[:, imin3:imax3]
return data_filtered
评论列表
文章目录
