def _denoise_band(X, wavelet, levels, alpha):
from var_est import variance
if alpha is None:
alpha = 2
decomp = pywt.wavedec2(X, wavelet, level=levels)
for i, all_coeff in enumerate(decomp[1:]):
minvar = np.empty(all_coeff[0].shape, dtype=float)
minvar.fill(np.inf)
# Handle horizontal, vertical and diagonal coefficients
for coeff in all_coeff:
for win_size in (3, 5, 7, 9):
var = variance(coeff, win_size)
mask = (var < minvar)
minvar[mask] = var[mask]
# Wiener estimator
coeff *= (minvar / (minvar + alpha))
rec = pywt.waverec2(decomp, wavelet)
rows, cols = X.shape
if X.shape != rec.shape:
rows_mod = rows % 2
cols_mod = cols % 2
return rec[rows_mod:, cols_mod:]
else:
return rec
def _denoise_band(X, wavelet, levels, alpha):
from var_est import variance
if alpha is None:
alpha = 2
decomp = pywt.wavedec2(X, wavelet, level=levels)
for i, all_coeff in enumerate(decomp[1:]):
minvar = np.empty(all_coeff[0].shape, dtype=float)
minvar.fill(np.inf)
# Handle horizontal, vertical and diagonal coefficients
for coeff in all_coeff:
for win_size in (3, 5, 7, 9):
var = variance(coeff, win_size)
mask = (var < minvar)
minvar[mask] = var[mask]
# Wiener estimator
coeff *= (minvar / (minvar + alpha))
rec = pywt.waverec2(decomp, wavelet)
rows, cols = X.shape
if X.shape != rec.shape:
rows_mod = rows % 2
cols_mod = cols % 2
return rec[rows_mod:, cols_mod:]
else:
return rec
def blend_images(base, texture, wavelet, level, mode='sp1', base_gain=None, texture_gain=None):
"""Blend loaded images at `level` of granularity using `wavelet`"""
base_data = image2array(base)
texture_data = image2array(texture)
output_data = []
# process color bands
for base_band, texture_band in zip(base_data, texture_data):
# multilevel dwt
base_band_coeffs = pywt.wavedec2(base_band, wavelet, mode, level)
texture_band_coeffs = pywt.wavedec2(texture_band, wavelet, mode, level)
# average coefficients of base image
output_band_coeffs = [base_band_coeffs[0]] # cA
del base_band_coeffs[0], texture_band_coeffs[0]
# blend details coefficints
for n, (base_band_details, texture_band_details) in enumerate(zip(base_band_coeffs, texture_band_coeffs)):
blended_details = []
for (base_detail, texture_detail) in zip(base_band_details, texture_band_details):
if base_gain is not None:
base_detail *= base_gain
if texture_gain is not None:
texture_detail *= texture_gain
# select coeffs with greater energy
blended = numpy.where(abs(base_detail) > abs(texture_detail), base_detail, texture_detail)
blended_details.append(blended)
base_band_coeffs[n] = texture_band_coeffs[n] = None
output_band_coeffs.append(blended_details)
# multilevel idwt
new_band = pywt.waverec2(output_band_coeffs, wavelet, mode)
output_data.append(new_band)
del new_band, base_band_coeffs, texture_band_coeffs
del base_data, texture_data
output_data = numpy.array(output_data)
return array2image(output_data, base.mode)
