def nrl(x, a=1, c=220):
"""
Lin-log style remap.
Parameters
----------
x : numpy.ndarray
data to remap
a : float
scale factor of 99th percentile for input "knee"
c : float
output "knee" in lin-log curve
Returns
-------
numpy.ndarray
"""
x = np.abs(x)
xmin = np.min(x)
p99 = prctile(x[np.isfinite(x)], 99)
b = (255 - c) / np.log10((np.max(x) - xmin) / ((a * p99) - xmin))
out = np.zeros_like(x, np.uint8)
linear_region = (x <= a * p99)
out[linear_region] = (x[linear_region] - xmin) * c / ((a * p99) - xmin)
out[np.logical_not(linear_region)] = c + (b * np.log10(
(x[np.logical_not(linear_region)] - xmin) / ((a * p99) - xmin)))
return out
def nrl(x, a=1., c=220.):
"""
Lin-log style remap.
Parameters
----------
x : numpy.ndarray
data to remap
a : float
scale factor of 99th percentile for input "knee"
c : float
output "knee" in lin-log curve
Returns
-------
numpy.ndarray
"""
x = numpy.abs(x)
xmax = numpy.max(x)
xmin = numpy.min(x)
p99 = prctile(x[numpy.isfinite(x)], 99)
b = (255 - c) / (numpy.log10(xmax - xmin) * (a * p99 - xmin))
out = numpy.zeros_like(x, numpy.uint8)
linear_region = (x <= a * p99)
out[linear_region] = c * (x[linear_region] - xmin) / (a * p99 - xmin)
out[~linear_region] = c + b * numpy.log10(
(x[~linear_region] - xmin) / (a * p99 - xmin))
return out
def nrl(x, a=1, c=220):
"""Lin-log style remap.
Input parameters
x : data to remap
a : scale factor of 99th percentile for input "knee"
c : output "knee" in lin-log curve
"""
# We include this import inside the function since it is only used here, and
# since we don't want importing of this module to fail if the user does not
# have scipy.
from scipy.stats import scoreatpercentile as prctile
x = abs(x)
xmin = np.min(x)
p99 = prctile(x[np.isfinite(x)], 99)
b = (255 - c) / np.log10((np.max(x) - xmin) / ((a * p99) - xmin))
out = np.zeros_like(x, np.uint8)
linear_region = (x <= a * p99)
out[linear_region] = (x[linear_region] - xmin) * c / ((a * p99) - xmin)
out[np.logical_not(linear_region)] = c + (b * np.log10(
(x[np.logical_not(linear_region)] - xmin) / ((a * p99) - xmin)))
return out
def _nrl_stats(amplitude):
"""
Calculate the statistiucs for input into the nrl remap.
Parameters
----------
amplitude : numpy.ndarray
The amplitude array, assumed real valued.
Returns
-------
tuple
Of the form `(minimum, maximum, 99th percentile)
"""
finite_mask = numpy.isfinite(amplitude)
if numpy.any(finite_mask):
return numpy.min(amplitude[finite_mask]), numpy.max(
amplitude[finite_mask]), prctile(amplitude[finite_mask], 99)
else:
return 0, 0, 0