How to normalize an image in OpenCV Python?

Image normalization in OpenCV rescales pixel values to a specific range, improving image processing and machine learning model performance. The cv2.normalize() function provides various normalization techniques to transform pixel intensities.

Syntax

The cv2.normalize() function accepts the following parameters ?

cv2.normalize(src, dst, alpha, beta, norm_type, dtype, mask)

Parameters

• src Input image array

• dst Output array of the same size as src

• alpha Lower norm value for range normalization

• beta Upper norm value for range normalization

• norm_type Normalization type (NORM_MINMAX, NORM_L2, etc.)

• dtype Data type of output array

• mask Optional operation mask

Example 1: Grayscale Image Normalization

This example normalizes a grayscale image to the range [0,1] using min-max normalization ?

import cv2
import numpy as np

# Create a sample grayscale image with different intensity values
img = np.array([[37, 45, 55, 120],
                [80, 90, 100, 150],
                [200, 220, 240, 255]], dtype=np.uint8)

print("Image data before Normalize:")
print(img)
print("Min value:", img.min(), "Max value:", img.max())

# Normalize the image to range [0, 1]
img_normalized = cv2.normalize(img, None, 0, 1.0, cv2.NORM_MINMAX, dtype=cv2.CV_32F)

print("\nImage data after Normalize:")
print(img_normalized)
print("Min value:", img_normalized.min(), "Max value:", img_normalized.max())

Image data before Normalize:
[[ 37  45  55 120]
 [ 80  90 100 150]
 [200 220 240 255]]
Min value: 37 Max value: 255

Image data after Normalize:
[[0.         0.03669725 0.08256881 0.38073394]
 [0.19724771 0.24311927 0.28899083 0.51834863]
 [0.74770642 0.8440367  0.93119266 1.        ]]
Min value: 0.0 Max value: 1.0

Example 2: Binary Image Normalization

This example creates a binary image and normalizes it to [0,1] range ?

import cv2
import numpy as np

# Create a sample grayscale image
img = np.array([[37, 45, 155, 120],
                [80, 190, 100, 150],
                [200, 220, 40, 255]], dtype=np.uint8)

print("Original image data:")
print(img)

# Apply threshold to create a binary image
ret, thresh = cv2.threshold(img, 140, 255, cv2.THRESH_BINARY)
print("\nImage data after Thresholding:")
print(thresh)

# Normalize the binary image
img_normalized = cv2.normalize(thresh, None, 0, 1.0, cv2.NORM_MINMAX, dtype=cv2.CV_32F)

print("\nImage data after Normalize:")
print(img_normalized)

Original image data:
[[ 37  45 155 120]
 [ 80 190 100 150]
 [200 220  40 255]]

Image data after Thresholding:
[[  0   0 255   0]
 [  0 255   0 255]
 [255 255   0 255]]

Image data after Normalize:
[[0. 0. 1. 0.]
 [0. 1. 0. 1.]
 [1. 1. 0. 1.]]

Normalization Types

OpenCV provides different normalization types ?

Example 3: Different Normalization Types

import cv2
import numpy as np

# Create sample data
data = np.array([3, 4, 5], dtype=np.float32)
print("Original data:", data)

# L2 normalization
l2_norm = cv2.normalize(data, None, 1.0, 0, cv2.NORM_L2)
print("L2 normalized:", l2_norm)

# L1 normalization  
l1_norm = cv2.normalize(data, None, 1.0, 0, cv2.NORM_L1)
print("L1 normalized:", l1_norm)

# Min-Max normalization to [0, 255]
minmax_norm = cv2.normalize(data, None, 0, 255, cv2.NORM_MINMAX)
print("Min-Max normalized:", minmax_norm)

Original data: [3. 4. 5.]
L2 normalized: [0.42426407 0.5656854  0.7071068 ]
L1 normalized: [0.25 0.33333334 0.41666667]
Min-Max normalized: [  0. 127.5 255. ]

Conclusion

Use cv2.normalize() with NORM_MINMAX for scaling images to specific ranges. Choose L1/L2 normalization for feature vectors in machine learning applications.