CODEWITHSUNDEEP
pythonnumpyopencv
TXN_747
TXN_747

Reputation: 79

Convert to grayscale using the max value of a channel

I'm trying to convert a color image to grayscale, but I would like to use the max value of the channels as the grayscale value. Given the sample image:

[[[0,100,200] [25,55,105] [11,22,33]],
 [[0,1,1] [222,111,0] [255,2,20]], 
 [[0,17,0] [0,0,0] [88,99,77]]]

I would like the resulting grayscale image to be

[[[200] [105] [33]], 
 [[1] [222] [255]],
 [[17] [0] [99]]]

Is there a better/faster method than splitting the channels and iterating through the image?

    def get_max_grayscale_image(img_mtx):
        img_b, img_g, img_r = cv2.split(img_mtx.copy())
        height, width = img_b.shape[0:2]
        gray_max_image = np.zeros((height, width), np.uint8)
        for i in range(height):
            for j in range(width):
                gray_max_image [i, j] = max(img_b[i, j], img_g[i, j], img_r [i, j])
        return gray_max_image

Upvotes: 1

Views: 61

Answers (1)

Mark
Mark

Reputation: 92461

You can use ndarray.max() and pass it the axis you want. Setting keepdims to True will keep this as a two dimensional list. If you want it flattened, you can leave that off or set it to False:

import numpy as np

im = np.array([
    [0,100,200], 
    [25,55,105],
    [11,22,33],
    [0,1,1],
    [222,111,0],
    [255,2,20],
    [0,17,0],
    [0,0,0],
    [88,99,77]
])

im.max(axis=1, keepdims=True)

This will produce a new array:

array([[200],
       [105],
       [ 33],
       [  1],
       [222],
       [255],
       [ 17],
       [  0],
       [ 99]])

Upvotes: 2

Related Questions