'''Code to convert RGB to other color spaces
'''
#
# Color transformations from
# /auto/k1/shinji/matlab/colorspace/colorspace2.m
# /auto/k1/shinji/matlab/strflab_adds/preprocColorSpace.m
#
# Code originally by Pascal Getreuer (https://getreuer.info/) 2005-2006:
# https://www.mathworks.com/matlabcentral/fileexchange/28790-colorspace-transformations
#
# Only a few transformations were converted to python.
#
# Anwar O. Nunez-Elizalde (2016).
#
# Updates:
#
#
import numpy as np
[docs]def rgb2lab(image):
'''Convert RGB to CIE LAB color space **in-place**
Parameters
----------
image : 3D numpy float array
Array must be in [0,1] range. Last
dimension corresponds to RGB channels.
Returns
-------
LAB : 3D numpy float array
The CIE LAB representation in the image.
'''
WhitePoint = np.asarray([0.95047,1.0,1.08883])
# Convert image to XYZ
image = rgb2xyz(image)
# % Convert XYZ to CIE L*a*b*
X = image[:,:,0]/WhitePoint[0]
Y = image[:,:,1]/WhitePoint[1]
Z = image[:,:,2]/WhitePoint[2]
fX = _ff(X)
fY = _ff(Y)
fZ = _ff(Z)
image[:,:,0] = 116.0*fY - 16 # L*
image[:,:,1] = 500.0*(fX - fY) # a*
image[:,:,2] = 200.0*(fY - fZ) # b*
return image
[docs]def rgb2lch(image):
'''Convert RGB to CIE LCH color space **in-place**
Parameters
----------
image : 3D numpy float array
Array must be in [0,1] range. Last
dimension corresponds to RGB channels.
Returns
-------
LCH : 3D numpy float array
The CIE LCH representation in the image.
'''
image = rgb2lab(image)
H = np.arctan2(image[:,:,2],image[:,:,1])
H = H*180/np.pi + 360*(H < 0.0)
image[:,:,1] = np.sqrt(image[:,:,1]**2 + image[:,:,2]**2) # C
image[:,:,2] = H # H
return image
[docs]def rgb2xyz(image):
'''Convert RGB to CIE XYZ color space **in-place**
Parameters
----------
image : 3D numpy float array
Array must be in [0,1] range. Last
dimension corresponds to RGB channels.
Returns
-------
LAB : 3D numpy float array
The CIE XYZ representation in the image.
'''
# % Undo gamma correction
R = _invgammacorrection(image[:,:,0])
G = _invgammacorrection(image[:,:,1])
B = _invgammacorrection(image[:,:,2])
# % Convert RGB to XYZ
xyz_from_rgb = np.array([[0.412453, 0.357580, 0.180423],
[0.212671, 0.715160, 0.072169],
[0.019334, 0.119193, 0.950227]])
T = xyz_from_rgb.T.ravel()
image[:,:,0] = T[0]*R + T[3]*G + T[6]*B # X
image[:,:,1] = T[1]*R + T[4]*G + T[7]*B # Y
image[:,:,2] = T[2]*R + T[5]*G + T[8]*B # Z
return image
def _ff(Y):
'''Obscure helper function
'''
fY = np.real(Y**(1./3.))
idx = Y < 0.008856
fY[idx] = Y[idx]*(7.787) + (4./29.)
return fY
def _invgammacorrection(Rp):
'''
'''
R = np.real(((Rp + 0.055)/1.055)**(2.4))
idx = Rp < 0.04045
R[idx] = Rp[idx] / 12.92
return R
def _gamma_correct(image, gamma=1.0):
"""Do not use unless you know what you're doing
"""
if gamma != 1.0:
if not (0 <= image.min() and 1 >= image.max()):
image /= 255.
image = image**gamma
return image
if __name__ == '__main__':
pass