Convert Gimp "Color Enhance" algorithm to python

Question

I am trying to code the same effect as "Color Enhance" in Gimp. Here is the doc: https://docs.gimp.org/2.8/en/plug-in-color-enhance.html

Basically it says : It does this by converting the colors to HSV space, measuring the range of saturation values across the image, then stretching this range to be as large as possible, and finally converting the colors back to RGB. It is similar to Stretch Contrast, except that it works in the HSV color space, so it preserves the hue.

I tried to implement the algorithm that way:

def enhanced(hsv):
    h, s, v = cv2.split(hsv)
    s_equ = cv2.equalizeHist(s)

    hsv_image = cv2.merge([h, s_equ, v])
    out = cv2.cvtColor(hsv_image, cv2.COLOR_HSV2BGR)
    cv2.imwrite("out.jpg", out)


img2 = cv2.imread("IMG_0233.jpg")

hsv2 = cv2.cvtColor(img2, cv2.COLOR_BGR2HSV)

enhanced(hsv2)

But I do not have the same result as Gimp:

Input image:

Result of Gimp color Enhance:

Result with my script:

Is there anything that I am missing? Many thanks.

Answer 1

Possibly one (likely Gimp) is working on "linear light" (likely if you are using the GEGL, non-legacy version of "Color enhance") while the other is using gamma-compressed values. In Gimp you can use the Pointer dialog to have the HSV value of pixels, and compare to what CV produces.

The source of the Gimp 2.10 color-enhance is here:

https://gitlab.gnome.org/GNOME/gegl/-/blob/master/operations/common/color-enhance.c

The comments state:

 * This operation is inspired from the color enhance gimp plugin.
 * It alters the chroma component to cover maximum possible range,
 * keeping hue and lightness untouched.