How to Apply Low Pass Filter to Image for Smoothing

To apply a low pass filter to an image, you typically use a smoothing technique like a Gaussian blur or averaging filter that removes high-frequency details. In Python, libraries like OpenCV or scipy.ndimage provide functions such as cv2.GaussianBlur() to apply this filter easily.

Syntax

The common syntax to apply a low pass filter using OpenCV's Gaussian blur is:

cv2.GaussianBlur(src, ksize, sigmaX)

• src: Input image.
• ksize: Kernel size (width, height), must be positive and odd.
• sigmaX: Gaussian kernel standard deviation in X direction.

This smooths the image by averaging pixel values weighted by a Gaussian function, reducing sharp edges and noise.

python

import cv2

# Apply Gaussian blur
blurred_image = cv2.GaussianBlur(src=image, ksize=(5,5), sigmaX=0)

Example

This example loads an image, applies a Gaussian low pass filter, and shows the original and filtered images side by side.

python

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

# Load image in grayscale
image = cv2.imread('example.jpg', cv2.IMREAD_GRAYSCALE)

# Apply Gaussian blur (low pass filter)
blurred = cv2.GaussianBlur(image, (7,7), sigmaX=1.5)

# Plot original and blurred images
plt.subplot(1,2,1)
plt.title('Original Image')
plt.imshow(image, cmap='gray')
plt.axis('off')

plt.subplot(1,2,2)
plt.title('Low Pass Filtered Image')
plt.imshow(blurred, cmap='gray')
plt.axis('off')

plt.show()

Output

A window showing two images side by side: the original grayscale image and the smoothed image with reduced noise and details.

Common Pitfalls

• Using an even kernel size for Gaussian blur will cause an error; always use odd numbers like (3,3), (5,5).
• Setting sigmaX to 0 lets OpenCV calculate it automatically, but setting it too high can over-blur the image.
• Applying a low pass filter to a color image requires handling each channel or converting to grayscale first.
• Not normalizing or converting image data types can cause unexpected results.

python

import cv2

# Wrong: even kernel size causes error
# blurred = cv2.GaussianBlur(image, (4,4), 0)  # This will fail

# Right: odd kernel size
blurred = cv2.GaussianBlur(image, (5,5), 0)

Quick Reference

Parameter	Description	Example Values
src	Input image to filter	Grayscale or color image array
ksize	Kernel size (width, height), odd integers	(3,3), (5,5), (7,7)
sigmaX	Gaussian kernel standard deviation in X	0 (auto), 1.0, 1.5
Output	Filtered image with reduced high-frequency details	Smoothed image array

Key Takeaways

Use Gaussian blur with an odd-sized kernel to apply a low pass filter to images.

Low pass filters smooth images by reducing sharp edges and noise.

Set sigmaX carefully to control the amount of smoothing.

Always check image type and channels before filtering.

Visualize results to confirm the filter effect.