How to Correct Skewed Text in Images Using OpenCV: A Step‑by‑Step Guide

Assume we have an image where the text is rotated by an unknown angle. To correct the orientation we follow four steps.

Step 1: Load image and binarize

img = cv.imread('img/imageTextR.png')
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
ret, thresh = cv.threshold(gray, 0, 255, cv.THRESH_BINARY_INV | cv.THRESH_OTSU)

We invert the binary image so that the text becomes white on a black background.

Step 2: Compute the minimum‑area bounding box of the text region

coords = np.column_stack(np.where(thresh > 0))
angle = cv.minAreaRect(coords)[-1]

The cv.minAreaRect function returns a rectangle whose rotation matches the text rotation.

Step 3: Adjust the angle

if angle < -45:
    angle = -(90 + angle)
else:
    angle = -angle

If the angle is less than –45°, we add 90°; otherwise we simply negate it.

Step 4: Apply affine rotation

h, w = img.shape[:2]
center = (w // 2, h // 2)
M = cv.getRotationMatrix2D(center, angle, 1.0)
rotated = cv.warpAffine(img, M, (w, h), flags=cv.INTER_CUBIC, borderMode=cv.BORDER_REPLICATE)
cv.putText(rotated, f'Angle: {angle:.2f} degrees', (10, 30), cv.FONT_HERSHEY_SIMPLEX, 0.7, (0,0,255), 2)

The corrected image is displayed alongside the original.