Philipp Klaus
610954f800
The dissection of the color into red, black, & white now takes part in the HSV color space (instead of HLS before). Pillow's built-in capabilities to convert RGB → HSV are now used. While the creation of --red labels still takes 16 times longer than the simple black/white ones, it is now 90% less time than before this commit.
24 lines
672 B
Python
24 lines
672 B
Python
from PIL import Image
|
|
import colorsys
|
|
|
|
def filtered_hsv(im, filter_h, filter_s, filter_v, default_col=(255,255,255)):
|
|
""" https://stackoverflow.com/a/22237709/183995 """
|
|
|
|
hsv_im = im.convert('HSV')
|
|
H, S, V = 0, 1, 2
|
|
hsv = hsv_im.split()
|
|
mask_h = hsv[H].point(filter_h)
|
|
mask_s = hsv[S].point(filter_s)
|
|
mask_v = hsv[V].point(filter_v)
|
|
|
|
Mdat = []
|
|
for h, s, v in zip(mask_h.getdata(), mask_s.getdata(), mask_v.getdata()):
|
|
Mdat.append(255 if (h and s and v) else 0)
|
|
|
|
mask = mask_h
|
|
mask.putdata(Mdat)
|
|
|
|
filtered_im = Image.new("RGB", im.size, color=default_col)
|
|
filtered_im.paste(im, None, mask)
|
|
return filtered_im
|