how to change Lab color space to RGB 0-255

user_12931358 0
Viewed 361

I have Lab dataset: val_lab_2 like that:

[[ 7.803e+01  3.100e-01  1.382e+01]
 [ 6.697e+01 -7.400e+00  2.750e+01]
 [ 5.631e+01 -1.804e+01  1.599e+01]
 [ 6.701e+01  2.650e+00  2.913e+01]
 [ 6.564e+01  1.660e+00  2.540e+01]
 [ 3.537e+01  2.050e+01  3.784e+01]
 [ 4.178e+01  2.251e+01  4.438e+01]
 [ 6.129e+01  1.261e+01  5.934e+01]
 [ 4.269e+01  5.120e+00  4.995e+01]...]

I want to change it to RGB 0-255, so I use colormath package:

import numpy as np
import pandas as pd 

from colormath.color_objects import sRGBColor, XYZColor, LabColor
from colormath.color_conversions import convert_color

val_rgb_2 = []
for lab_list in val_lab_2:
       lab = LabColor(*[component for component in lab_list])
       rgb = convert_color(lab, sRGBColor)
       rgb_list = [255*color for color in rgb.get_value_tuple()]
       val_rgb_2.append(rgb_list)
val_rgb_2 = np.array(val_rgb_2)
print(val_rgb_2)

the result shows:

[[201.44003158 192.14717152 167.2643737 ]
 [163.51015463 166.15931613 112.52909259]
 [109.8451797  143.64817993 106.17872676]
 [181.45664244 160.44644464 110.27374654]
 [174.70091435 157.51328159 113.65978481]
 [122.18753543  69.26114552  19.79086107]
 [143.2650315   82.85139354  20.50673141]
 [187.60706038 138.48640317  31.06087929]...]

However, I think it is not correct, because I have a label, first few rows shows it should be:

natural white 100% wool,blue-violet flowers,blue flowers,flowers,whole plant,peels,peels,peels...

colormath computer-vision matplotlib opencv python
1 Answers

I suggest you to use this website to:

  • insert the LAB values
  • check the RGB values that the website gives you
  • check the color associated with the LAB/RGB you have insert

I have done a check and the conversion between the two matrix you wrote is correct.
For example let's try with the second element of the list above, which it should be labeled as blue-violet flowers:

[ 6.697e+01 -7.400e+00  2.750e+01]

enter image description here

The conversion is correct; I have some doubts regarding the labels.

user_92143570
Related