How to apply Perspective Transformations on an image using OpenCV Python?

In Perspective Transformation, straight lines remain straight even after the transformation. To apply a perspective transformation, we need a 3×3 perspective transformation matrix and four points on both the input and output images.

Key Functions

We use cv2.getPerspectiveTransform() to find the transformation matrix ?

M = cv2.getPerspectiveTransform(pts1, pts2)

Where:

• pts1 ? An array of four points on the input image

• pts2 ? An array of corresponding four points on the output image

To apply the transformation, we use cv2.warpPerspective() ?

dst = cv2.warpPerspective(img, M, (width, height))

Where:

• img ? The image to be transformed

• M ? Perspective transformation matrix

• (width, height) ? Dimensions of the output image

Basic Example

Let's apply perspective transformation to correct a skewed rectangular object ?

import cv2
import numpy as np

# Read the input image
img = cv2.imread('/python/opencv/images/warning_wall.jpg')

# Get image dimensions
rows, cols, ch = img.shape

# Define four points on the skewed rectangle in input image
pts1 = np.float32([[56,65], [368,52], [28,387], [389,390]])

# Define corresponding points for a perfect rectangle in output
pts2 = np.float32([[0,0], [300,0], [0,300], [300,300]])

# Calculate perspective transform matrix
M = cv2.getPerspectiveTransform(pts1, pts2)

# Apply perspective transformation
dst = cv2.warpPerspective(img, M, (300, 300))

# Display results
cv2.imshow('Original Image', img)
cv2.imshow('Transformed Image', dst)
cv2.waitKey(0)
cv2.destroyAllWindows()

Advanced Example with Matplotlib

This example shows side-by-side comparison using matplotlib ?

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

# Read image in grayscale
img = cv2.imread('/python/opencv/images/warning_wall.jpg', 0)

# Define source and destination points
pts1 = np.float32([[56,65], [368,52], [28,387], [389,390]])
pts2 = np.float32([[0,0], [400,0], [0,400], [400,400]])

# Get transformation matrix
M = cv2.getPerspectiveTransform(pts1, pts2)

# Apply transformation with custom output size
dst = cv2.warpPerspective(img, M, (400, 400))

# Display using matplotlib
plt.figure(figsize=(12, 5))
plt.subplot(121)
plt.imshow(img, cmap='gray')
plt.title('Original Image')
plt.axis('off')

plt.subplot(122)
plt.imshow(dst, cmap='gray')
plt.title('Perspective Corrected')
plt.axis('off')

plt.tight_layout()
plt.show()

Common Use Cases

Key Points

• Always define points in the same order (e.g., top-left, top-right, bottom-left, bottom-right)

• Use np.float32() for point arrays to ensure proper data type

• Output dimensions can be different from input dimensions

• The transformation preserves straight lines but changes angles and distances

Conclusion

Perspective transformation is essential for correcting geometric distortions in images. Use cv2.getPerspectiveTransform() to calculate the matrix and cv2.warpPerspective() to apply the transformation with your desired output dimensions.