How to Rotate Image Using OpenCV in Computer Vision

To rotate an image in OpenCV, use cv2.getRotationMatrix2D to create a rotation matrix and then apply cv2.warpAffine to perform the rotation. This method lets you specify the rotation center, angle in degrees, and scale factor.

Syntax

The rotation process in OpenCV involves two main steps:

• Get rotation matrix: cv2.getRotationMatrix2D(center, angle, scale) creates a matrix to rotate the image around a point.
• Apply rotation: cv2.warpAffine(image, rotation_matrix, output_size) applies the rotation matrix to the image.

Parameters explained:

• center: The point (x, y) around which to rotate the image, usually the image center.
• angle: Rotation angle in degrees. Positive values rotate counter-clockwise.
• scale: Scaling factor to resize the image during rotation (1.0 means no scaling).
• output_size: Size (width, height) of the output image.

python

rotation_matrix = cv2.getRotationMatrix2D(center, angle, scale)
rotated_image = cv2.warpAffine(image, rotation_matrix, (width, height))

Example

This example loads an image, rotates it 45 degrees around its center without scaling, and displays both original and rotated images.

python

import cv2

# Load image
image = cv2.imread('input.jpg')
height, width = image.shape[:2]

# Define rotation center
center = (width // 2, height // 2)
angle = 45  # degrees
scale = 1.0  # no scaling

# Get rotation matrix
rotation_matrix = cv2.getRotationMatrix2D(center, angle, scale)

# Rotate image
rotated_image = cv2.warpAffine(image, rotation_matrix, (width, height))

# Save rotated image
cv2.imwrite('rotated_output.jpg', rotated_image)

# Optional: Display images (requires GUI support)
# cv2.imshow('Original', image)
# cv2.imshow('Rotated', rotated_image)
# cv2.waitKey(0)
# cv2.destroyAllWindows()

Output

Creates a file 'rotated_output.jpg' with the image rotated 45 degrees around its center.

Common Pitfalls

• Wrong rotation center: Not using the image center can cause unexpected cropping or shifting.
• Output size mismatch: Using the original image size for output can crop rotated corners; consider adjusting size to fit rotated image.
• Angle direction confusion: Positive angles rotate counter-clockwise; negative angles rotate clockwise.
• Scaling issues: Forgetting to set scale to 1.0 can resize the image unintentionally.

python

import cv2

image = cv2.imread('input.jpg')
height, width = image.shape[:2]

# Wrong: rotation center at (0,0) causes shift
wrong_center = (0, 0)
rotation_matrix_wrong = cv2.getRotationMatrix2D(wrong_center, 45, 1.0)
rotated_wrong = cv2.warpAffine(image, rotation_matrix_wrong, (width, height))
cv2.imwrite('rotated_wrong.jpg', rotated_wrong)

# Right: rotation center at image center
center = (width // 2, height // 2)
rotation_matrix_right = cv2.getRotationMatrix2D(center, 45, 1.0)
rotated_right = cv2.warpAffine(image, rotation_matrix_right, (width, height))
cv2.imwrite('rotated_right.jpg', rotated_right)

Output

Creates 'rotated_wrong.jpg' with shifted image and 'rotated_right.jpg' correctly rotated around center.

Quick Reference

Tips for rotating images in OpenCV:

• Use cv2.getRotationMatrix2D with the image center for natural rotation.
• Set scale=1.0 to keep original size unless resizing is needed.
• Adjust output size if you want to avoid cropping rotated corners.
• Remember positive angles rotate counter-clockwise.

Key Takeaways

Use cv2.getRotationMatrix2D with the image center, angle, and scale to create the rotation matrix.

Apply cv2.warpAffine with the rotation matrix and output size to rotate the image.

Positive angles rotate counter-clockwise; negative angles rotate clockwise.

Ensure the rotation center is correct to avoid unwanted image shifts.

Adjust output size to prevent cropping of rotated image corners.