CODEWITHSUNDEEP
pythonopencv
Krivers
Krivers

Reputation: 2036

Convert to custom color space

I would like to ask you, how can I convert BGR image loaded by OpenCV (3.4.5) in python to the color space defined by the following formula:

enter image description here

I have following idea, buy its not correct.

import cv2
imageSource = cv2.imread("test.jpg")
A = np.array([
              [ 0.06,  0.63 ,  0.27],
              [ 0.3 ,  0.04 , -0.35],
              [ 0.34, -0.6  ,  0.17]
             ])
vis0 = cv2.multiply(imageSource, A)

Upvotes: 2

Views: 1491

Answers (1)

jedwards
jedwards

Reputation: 30200

You could do something like:

import cv2
import numpy as np

# This is just a simple helper function that takes a matrix, converts it to
# the BGR colorspace (if necessary), shows it with .imshow() and waits using
# .waitKey() -- feel free to ignore it.
def show(*, img_bgr=None, img_rgb=None):
    if img_bgr is None:
        img_bgr = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2BGR)
    cv2.imshow("title", img_bgr)
    cv2.waitKey()

transform = np.array([
    [ 0.06,  0.63 ,  0.27],
    [ 0.3 ,  0.04 , -0.35],
    [ 0.34, -0.6  ,  0.17]
])

img_bgr = cv2.imread("lenna.png",)
# The image will be in BGR order, we want RGB order
img_rgb = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2RGB)

# This does three things:
# - Transforms the pixels according to the transform matrix
# - Rounds the pixel values to integers
# - Coverts the datatype of the matrix to 'uint8' show .imshow() works
img_trans = np.rint(img_rgb.dot(transform.T)).astype('uint8')

show(img_bgr=img_bgr)
show(img_rgb=img_trans)

Which, from:

enter image description here

produces:

enter image description here

Note: If you were looking for something like:

enter image description here

Then drop the transpose (...dot(transform.T)... -> ...dot(transform)...)

Upvotes: 1

Related Questions