How to create a "bluescale" image-CodePudding

How to create an image manipulation effect similar to this website using opencv:

Using the following code:

import cv2

img = cv2.imread("./EZCUD.jpg")
img_blue = img.copy()
img_blue = cv2.cvtColor(img_blue, cv2.COLOR_BGR2HSV)
img_blue[:, :, 0] = 120
cv2.imwrite("./test.jpg", cv2.cvtColor(img_blue, cv2.COLOR_HSV2BGR))

Edit:

I managed to get a much better result by applying a gamma correction on the independent BGR channels after the HSV processing. I normalize the image to the range [0,1] and took square power in the RG channels and took the square root in the Blue channel.

This is the result:

With the following code:

import cv2
import numpy as np

img = cv2.imread("./EZCUD.jpg")
img_blue = img.copy()

img_blue = cv2.cvtColor(img_blue, cv2.COLOR_BGR2HSV)
img_blue[:, :, 0] = 120
img_blue = cv2.cvtColor(img_blue, cv2.COLOR_HSV2BGR)
img_blue = img_blue.astype("float")
img_blue /= 255
img_blue[:, :, 1:] = img_blue[:, :, 1:] ** 2
img_blue[:, :, 0] = img_blue[:, :, 0] ** (1 / 2)
img_blue *= 255

cv2.imwrite("./test.jpg", img_blue.astype("uint8"))

You can play around with these parameters until the result is closer to what you expected.

CodePudding user response：

You can try convert RGB to BGR by using cv2.cvtColor or apply colormap with cv2.applyColorMap

import cv2
img = cv2.imread("car.jpg")
# First method
img_bgr = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
# Second method
img_ocean = cv2.applyColorMap(img, cv2.COLORMAP_OCEAN)

Here are example outputs for img_bgr and img_ocean. Hope it helps.

Edited: You can also try customize your own ColorMap