One requirement is to convert the red part of an image to the blue part. Initially, one approach could be to solve this by swapping channels, but this may result in some mixed color transformations. For example, orange turning into purple (yellow + red || yellow-green + blue).
Another approach is color space transformation. When a problem cannot be solved within one space, it can be transformed to another space through coordinate transformation for solving, and then transformed back. In the other space, our operations will not impose unexpected operations on the original space.
Implementation code:
import cv2
import numpy as np
image = cv2.imread('C:/Users/admin/Desktop/test.png')
HSVImage = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
lowerRed1 = np.array([0, 64, 50])
upperRed1 = np.array([10, 255, 255])
lowerRed2 = np.array([160, 64, 50])
upperRed2 = np.array([180, 255, 255])
redMask1 = cv2.inRange(HSVImage, lowerRed1, upperRed1)
redMask2 = cv2.inRange(HSVImage, lowerRed2, upperRed2)
fullMask = redMask1 + redMask2
blueHue = 60
HSVImage[:, :, 0] = np.where(fullMask == 255, blueHue, HSVImage[:, :, 0])
finalImage = cv2.cvtColor(HSVImage, cv2.COLOR_HSV2BGR)
cv2.imwrite('C:/Users/admin/Desktop/out.png', finalImage)