Numpy array of a png file is rotated in matplotlib.pyplot

Question

I'm trying to scatter all white pixels of this gradient image in matplotlib.pyplot:

import numpy as np
from PIL import Image
import cv2
from matplotlib import pyplot

img = Image.open(
    "/root/.../aec.png").convert("L")

img = np.array(img)

kernel = np.ones((2, 2), np.uint8)
gradient = cv2.morphologyEx(img, cv2.MORPH_GRADIENT, kernel)

ox, oy = np.where(gradient == 255)

pyplot.plot(ox, oy, ".k")
pyplot.show()

The original picture (...) has a resolution of 2123x1269 and looks like this:

My graph in pyplot shows my gradient picture 270° rotated clockwise and I don't understand why.

I tried pyplot.plot(oy, ox, ".k") , then it's flipped to the x-axis compared to the original image. Rotating the original image with gradient = cv2.rotate(gradient, cv2.cv2.ROTATE_90_CLOCKWISE) gives me coordinates different from the pixel coordinates of my orginal image. ( The xy-pixel-coordinates have to be the ones of the gradient image. ) Also the resolution of 2123x1269 should remain and the program should run as fast as possible.

How can I display the pixel coordinates of the gradient image in matplotlib.pyplot correctly?

Answer 1

That is because origin in opencv is at the top-left. Try reverting y axis on pyplot and exchange x and y.

EDIT: Just use plt.imshow(), it is the right function to display image data.

Answer 2

For anyone who ever encounters this problem, this is my final code:

import numpy as np
from PIL import Image
import cv2
from matplotlib import pyplot

img = Image.open(
    "/root/.../aec.png").convert("L")

img = np.array(img)

kernel = np.ones((2, 2), np.uint8)
gradient = cv2.morphologyEx(img, cv2.MORPH_GRADIENT, kernel)

ox, oy = np.where(gradient == 255)

pyplot.plot(oy, ox, ".k")
pyplot.imshow(gradient)
pyplot.show()

Zoomed in version of the plot: correct plot , until now it servers my purpose.

Thanks @frab