How can I achieve PIL colorize functionality?

Question

Using PIL, I can transform an image's color by first converting it to grayscale and then applying the colorize transform. Is there a way to do the same with scikit-image?

The difference with e.g. the question at Color rotation in HSV using scikit-image is that there the black stays black while in PIL colorize function, I can define both where I want black and white mapped to.

Answer 1

I think you want something like this to avoid any dependency on PIL/Pillow:

#!/usr/bin/env python3

import numpy as np
from PIL import Image

def colorize(im,black,white):
    """Do equivalent of PIL's "colorize()" function"""
    # Pick up low and high end of the ranges for R, G and B
    Rlo, Glo, Blo = black
    Rhi, Ghi, Bhi = white

    # Make new, empty Red, Green and Blue channels which we'll fill & merge to RGB later
    R = np.zeros(im.shape, dtype=np.float)
    G = np.zeros(im.shape, dtype=np.float)
    B = np.zeros(im.shape, dtype=np.float)

    R = im/255 * (Rhi-Rlo) + Rlo
    G = im/255 * (Ghi-Glo) + Glo
    B = im/255 * (Bhi-Blo) + Blo

    return (np.dstack((R,G,B))).astype(np.uint8)


# Create black-white left-right gradient image, 256 pixels wide and 100 pixels tall
grad = np.repeat(np.arange(256,dtype=np.uint8).reshape(1,-1), 100, axis=0) 
Image.fromarray(grad).save('start.png')

# Colorize from green to magenta
result = colorize(grad, [0,255,0], [255,0,255])

# Save result - using PIL because I don't know skimage that well
Image.fromarray(result).save('result.png')

That will turn this:

into this:

---

Note that this is the equivalent of ImageMagick's -level-colors BLACK,WHITE operator which you can do in Terminal like this:

convert input.png -level-colors lime,magenta result.png

That converts this:

into this:

---

Keywords: Python, PIL, Pillow, image, image processing, colorize, colorise, colourise, colourize, level colors, skimage, scikit-image.