检查图像的 RGB 值是否满足没有循环的条件

Question

Say I have a coloured image img, as defined below. And I have range values for R, G, and B: R1, R2, B1, and so on...

Now I want to set all RGB values of an image to [255, 0, 0] if the following conditions satisfied [(R1 > R & R < R2) & (G1 > G & G < G2) & (B1 > B & B < B2)].

I can do this by looping to all RGB values of the image, but I don't want to do it that way. Is there a way that I can implement this one in the NumPyish way? Or what is the best way to implement this one?

img = cv2.imread(file)
R1 = 91
R2 = 150
G1 = 10
G2 = 100
B1 = 100
B2 = 150

Answer 1

You can do the colour comparison for each colour channel at once using numpy like img[:,:,2] > R1 for the red channel. You can combine all these together like so.

# opencv stores colour channels in BGR order by default normally
img[(img[:,:,0] > B1) &(img[:,:,0] < B2) & (img[:,:,1] > G1) & (img[:,:,1] < G2) & (img[:,:,2] > R1) & (img[:,:,2] < R2),:] = [0,0,255]

Answer 2

You can put the elements R1,R2,...,G1,G2,...,B1,B2,... into an array and check for the conditions in the following way:

import numpy as np
R = 90 # example
array=np.array([[R1,R2],[G1,G2],[B1,B2]])

And using the numpy where function, you can specify the condition, eg <R and the operations if the condition is satisfied or not satisfied.

np.where(array[:,0]<R, 255, array[:,0]) # for R1,G1,B1
np.where(array[:,1]<R, 0, array[:,0]) # for R2,G2,B2 and so on

You can modify from here.

np.where(array[:,0]<R, 255, array[:,0]) goes through elements and if an element is less than R replaces the element with 255 and if not then does not modify the element.

Answer 3

Just create a function and apply it over your elements. Might still use a for loop internally. Images are BGR format in opencv.

func = lambda x: [0, 0, 255] if (R1 > x[2] & x[2] < R2) & (G1 > x[1] & x[1] < G2) & (B1 > x[0] & x[0] < B2) else x
img2 = numpy.apply_along_axis(func, 0, img)