Perform image processing tasks using SciPy's ndimage module

Last update on March 27 2025 09:14:12 (UTC/GMT +8 hours)

15. Image Processing with SciPy ndimage

Write a NumPy program to create a NumPy array and use SciPy's ndimage module to perform image processing tasks, such as filtering and edge detection.

Sample Solution:

Python Code:

import numpy as np  # Import NumPy library
from scipy import ndimage  # Import ndimage module from SciPy
import matplotlib.pyplot as plt  # Import matplotlib for plotting

# Create a NumPy array representing a simple 2D image (e.g., a checkerboard pattern)
def create_checkerboard(size=8):
    """Create a checkerboard pattern."""
    row_even = [1, 0] * size
    row_odd = [0, 1] * size
    board = []
    for i in range(size):
        if i % 2 == 0:
            board.append(row_even)
        else:
            board.append(row_odd)
    return np.array(board)

# Generate a checkerboard image
image = create_checkerboard(8)

# Apply Gaussian filter to the image
filtered_image = ndimage.gaussian_filter(image, sigma=1)

# Apply Sobel filter for edge detection
sobel_h = ndimage.sobel(image, axis=0)  # Horizontal edges
sobel_v = ndimage.sobel(image, axis=1)  # Vertical edges
sobel_combined = np.hypot(sobel_h, sobel_v)  # Combine both directions

# Plot the original and processed images
plt.figure(figsize=(12, 8))

plt.subplot(2, 2, 1)
plt.title("Original Image")
plt.imshow(image, cmap='gray', interpolation='nearest')
plt.axis('off')

plt.subplot(2, 2, 2)
plt.title("Gaussian Filtered Image")
plt.imshow(filtered_image, cmap='gray', interpolation='nearest')
plt.axis('off')

plt.subplot(2, 2, 3)
plt.title("Sobel Edge Detection (Horizontal)")
plt.imshow(sobel_h, cmap='gray', interpolation='nearest')
plt.axis('off')

plt.subplot(2, 2, 4)
plt.title("Sobel Edge Detection (Combined)")
plt.imshow(sobel_combined, cmap='gray', interpolation='nearest')
plt.axis('off')

# Display the plots
plt.tight_layout()
plt.show()

Output:

Explanation:

• Import Libraries:
• Import the NumPy library for creating and manipulating arrays.
• Import the "ndimage" module from SciPy for image processing tasks.
• Import Matplotlib for plotting the images.
• Create a Checkerboard Pattern:
• Define a function create_checkerboard(size=8) to generate a checkerboard pattern as a 2D NumPy array.
• Use this function to create a sample image for processing.
• Apply Gaussian Filter:
• Use ndimage.gaussian_filter(image, sigma=1) to apply a Gaussian filter to the image. The sigma parameter controls the extent of the blurring.
• Apply Sobel Filter for Edge Detection:
• Use ndimage.sobel(image, axis=0) to detect horizontal edges in the image.
• Use ndimage.sobel(image, axis=1) to detect vertical edges in the image.
• Combine the horizontal and vertical edges using np.hypot(sobel_h, sobel_v) to create a comprehensive edge detection result.
• Plot the Original and Processed Images:
• Use Matplotlib to plot the original image, the Gaussian filtered image, the Sobel horizontal edges, and the combined Sobel edges in a 2x2 grid.
• Display the Plots:
• Use plt.tight_layout() to adjust the layout and plt.show() to display the plots.

For more Practice: Solve these Related Problems:

• Write a Numpy program to apply a Gaussian filter on an image array using SciPy's ndimage and then perform edge detection.
• Write a Numpy program to use SciPy's ndimage to perform image segmentation on a noisy image.
• Write a Numpy program to rotate an image by a specified angle using SciPy's ndimage.rotate and compare with manual interpolation.
• Write a Numpy program to apply a median filter on an image array with SciPy's ndimage.median_filter and assess the noise reduction.

Python-Numpy Code Editor:

Previous: Perform optimization using SciPy's optimize module to find function minimum.
Next: Generate and analyze synthetic data with NumPy and SciPy.