How could I implement a centered shear an image with opencv

Question

When I use warpAffine to shear an image:

M2 = np.float32([[1, 0, 0], [0.2, 1, 0]])
aff2 = cv2.warpAffine(im, M2, (W, H))

I obtain an image that is not sheared around the image center. I can see black triangular areas on one side of the image, and the other side does not have black areas.

How could I let the image be sheared symmetricly ?

Answer 1

You have to adjust your translation parameters (3rd column) to center your image. ie you have to translate half the width and height multiplied by a factor.

For example

M2 = np.float32([[1, 0, 0], [0.2, 1, 0]])
M2[0,2] = -M2[0,1] * W/2
M2[1,2] = -M2[1,0] * H/2
aff2 = cv2.warpAffine(im, M2, (W, H))

Before

After

Full code

import cv2
import numpy as np
import matplotlib.pyplot as plt

im = np.ones((100,100))
H, W = im.shape

M2 = np.float32([[1, 0, 0], [0.2, 1, 0]])
M2[0,2] = -M2[0,1] * W/2
M2[1,2] = -M2[1,0] * H/2
aff2 = cv2.warpAffine(im, M2, (W, H))

plt.imshow(aff2, cmap="gray")