github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-08-22 14:24:40 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

Formatting

Browse source
This commit is contained in:
Jeremy Hummel 2019-08-10 00:47:48 -07:00
commit 62c7be73b6
1 changed files with 88 additions and 84 deletions
Download patch file Download diff file Expand all files Collapse all files

12
imagelib/color_transfer.py
View file

@ -1,6 +1,7 @@
import numpy as np import numpy as np
import cv2 import cv2
def reinhard_color_transfer(target, source, clip=False, preserve_paper=False, source_mask=None, target_mask=None): def reinhard_color_transfer(target, source, clip=False, preserve_paper=False, source_mask=None, target_mask=None):
""" """
Transfers the color distribution from the source to the target Transfers the color distribution from the source to the target
@ -35,7 +36,6 @@ def reinhard_color_transfer(target, source, clip=False, preserve_paper=False, so
OpenCV image (w, h, 3) NumPy array (uint8) OpenCV image (w, h, 3) NumPy array (uint8)
""" """
# convert the images from the RGB to L*ab* color space, being # convert the images from the RGB to L*ab* color space, being
# sure to utilizing the floating point data type (note: OpenCV # sure to utilizing the floating point data type (note: OpenCV
# expects floats to be 32-bit, so use that instead of 64-bit) # expects floats to be 32-bit, so use that instead of 64-bit)
@ -85,13 +85,14 @@ def reinhard_color_transfer(target, source, clip=False, preserve_paper=False, so
# return the color transferred image # return the color transferred image
return transfer return transfer
def linear_color_transfer(target_img, source_img, mode='pca', eps=1e-5): def linear_color_transfer(target_img, source_img, mode='pca', eps=1e-5):
''' """
Matches the colour distribution of the target image to that of the source image Matches the colour distribution of the target image to that of the source image
using a linear transform. using a linear transform.
Images are expected to be of form (w,h,c) and float in [0,1]. Images are expected to be of form (w,h,c) and float in [0,1].
Modes are chol, pca or sym for different choices of basis. Modes are chol, pca or sym for different choices of basis.
''' """
mu_t = target_img.mean(0).mean(0) mu_t = target_img.mean(0).mean(0)
t = target_img - mu_t t = target_img - mu_t
t = t.transpose(2, 0, 1).reshape(3, -1) t = t.transpose(2, 0, 1).reshape(3, -1)
@ -123,6 +124,7 @@ def linear_color_transfer(target_img, source_img, mode='pca', eps=1e-5):
matched_img[matched_img < 0] = 0 matched_img[matched_img < 0] = 0
return matched_img return matched_img
def lab_image_stats(image): def lab_image_stats(image):
# compute the mean and standard deviation of each channel # compute the mean and standard deviation of each channel
(l, a, b) = cv2.split(image) (l, a, b) = cv2.split(image)
@ -133,6 +135,7 @@ def lab_image_stats(image):
# return the color statistics # return the color statistics
return (lMean, lStd, aMean, aStd, bMean, bStd) return (lMean, lStd, aMean, aStd, bMean, bStd)
def _scale_array(arr, clip=True): def _scale_array(arr, clip=True):
if clip: if clip:
return np.clip(arr, 0, 255) return np.clip(arr, 0, 255)
@ -146,6 +149,7 @@ def _scale_array(arr, clip=True):
return arr return arr
def channel_hist_match(source, template, hist_match_threshold=255, mask=None): def channel_hist_match(source, template, hist_match_threshold=255, mask=None):
# Code borrowed from: # Code borrowed from:
# https://stackoverflow.com/questions/32655686/histogram-matching-of-two-images-in-python-2-x # https://stackoverflow.com/questions/32655686/histogram-matching-of-two-images-in-python-2-x
@ -176,6 +180,7 @@ def channel_hist_match(source, template, hist_match_threshold=255, mask=None):
return interp_t_values[bin_idx].reshape(oldshape) return interp_t_values[bin_idx].reshape(oldshape)
def color_hist_match(src_im, tar_im, hist_match_threshold=255): def color_hist_match(src_im, tar_im, hist_match_threshold=255):
h, w, c = src_im.shape h, w, c = src_im.shape
matched_R = channel_hist_match(src_im[:, :, 0], tar_im[:, :, 0], hist_match_threshold, None) matched_R = channel_hist_match(src_im[:, :, 0], tar_im[:, :, 0], hist_match_threshold, None)
@ -186,6 +191,5 @@ def color_hist_match(src_im, tar_im, hist_match_threshold=255):
for i in range(3, c): for i in range(3, c):
to_stack += (src_im[:, :, i],) to_stack += (src_im[:, :, i],)
matched = np.stack(to_stack, axis=-1).astype(src_im.dtype) matched = np.stack(to_stack, axis=-1).astype(src_im.dtype)
return matched return matched