note.nkmk.me

Convert BGR and RGB with Python, OpenCV (cvtColor)

Posted: | Tags: Python, OpenCV, Image Processing

When the image file is read with the OpenCV function imread(), the order of colors is BGR (blue, green, red). On the other hand, in Pillow, the order of colors is assumed to be RGB (red, green, blue).

Therefore, if you want to use both the Pillow function and the OpenCV function, you need to convert BGR and RGB.

You can use the OpenCV function cvtColor() or simply change the order of ndarray.

This article describes the following contents.

  • OpenCV is BGR, Pillow is RGB
  • Convert BGR and RGB with OpenCV function cvtColor()
  • Convert BGR and RGB without using cvtColor()

OpenCV is BGR, Pillow is RGB

When reading a color image file, OpenCV imread() reads as a NumPy array ndarray of row (height) x column (width) x color (3). The order of color is BGR (blue, green, red).

The OpenCV function imwrite() that saves an image assumes that the order of colors is BGR, so it is saved as a correct image.

import cv2
import numpy as np
from PIL import Image

im_cv = cv2.imread('data/src/lena.jpg')

cv2.imwrite('data/dst/lena_bgr_cv.jpg', im_cv)
source: opencv_cvtcolor_bgr_rgb.py

bgr opencv imwrite

When performing image processing with Pillow, you can convert ndarray to a PIL.Image object with Image.fromarray(), but in Pillow the color order assumes RGB (red, green, blue).

Therefore, if the ndarray of the image read by OpenCV imread() is converted to a PIL.Image object and saved, the image with the wrong color is saved.

pil_img = Image.fromarray(im_cv)
pil_img.save('data/dst/lena_bgr_pillow.jpg')
source: opencv_cvtcolor_bgr_rgb.py

bgr pillow save

If you want to convert ndarray and PIL.Image objects to use both Pillow and OpenCV functions, you need to convert BGR and RGB.

Convert BGR and RGB with OpenCV function cvtColor()

Various color spaces such as RGB, BGR, HSV can be mutually converted using OpenCV function cvtColor().

dst = cv2.cvtColor(src, code)

Refer to the following document for the value to be specified for the parameter code.

When code is cv2.COLOR_BGR2RGB, BGR is converted to RGB.

When converted to RGB, it will be saved as a correct image even if it is saved after being converted to a PIL.Image object.

im_rgb = cv2.cvtColor(im_cv, cv2.COLOR_BGR2RGB)

Image.fromarray(im_rgb).save('data/dst/lena_rgb_pillow.jpg')
source: opencv_cvtcolor_bgr_rgb.py

rgb pillow save

When converted to RGB and saved with OpenCV imwrite(), it will be an incorrect color image.

cv2.imwrite('data/dst/lena_rgb_cv.jpg', im_rgb)
source: opencv_cvtcolor_bgr_rgb.py

rgb opencv imwrite

The parameter code when converting from RGB to BGR is cv2.COLOR_RGB2BGR. Use this when reading an image file as a PIL.Image, convert it to ndarray, and save it using OpenCV imwrite().

im_pillow = np.array(Image.open('data/src/lena.jpg'))

im_bgr = cv2.cvtColor(im_pillow, cv2.COLOR_RGB2BGR)

cv2.imwrite('data/dst/lena_bgr_cv_2.jpg', im_bgr)
source: opencv_cvtcolor_bgr_rgb.py

Convert BGR and RGB without using cvtColor()

Converting BGR and RGB can be realized without using cvtColor().

There are several ways, for example, as follows:

im_bgr = cv2.imread('data/src/lena.jpg')

im_rgb = im_bgr[:, :, [2, 1, 0]]
Image.fromarray(im_rgb).save('data/dst/lena_swap.jpg')

im_rgb = im_bgr[:, :, ::-1]
Image.fromarray(im_rgb).save('data/dst/lena_swap_2.jpg')
source: opencv_cvtcolor_bgr_rgb.py

Related Categories

Related Articles