A Beginner's Guide to Python OpenCV Morphological Operations (Easy to Understand!)

What are Morphological Operations?¶

In image processing, morphological operations are a class of shape-based processing methods. The core idea is to interact with the image through a Structuring Element (Kernel) to modify the shape characteristics of objects in the image. Morphological operations are primarily used on binary images (black-and-white images) and achieve effects like noise removal, connecting broken objects, and filling holes through basic operations such as “dilation” and “erosion”.

Basic Types of Morphological Operations¶

Morphological operations have several fundamental forms, with the four core types being:

1. Erosion¶

Principle: Erosion can be understood as “shrinking” the bright regions (white parts) of the image or “expanding” the dark regions (black parts). It scans the image using a structuring element and sets all pixels within the area covered by the structuring element to the minimum value (if any pixel in the covered area is 0, the result is 0).

Effect: Removes small bright spots (noise) but causes the object edges to “shrink”, potentially making the object smaller or even disappear.

2. Dilation¶

Principle: Opposite to erosion, dilation “expands” the bright regions of the image. It scans the image using a structuring element and sets all pixels within the covered area to the maximum value (if any pixel in the covered area is 255, the result is 255).

Effect: Fills small dark holes (cavities) in the image, expands object edges, and may connect broken objects.

3. Opening¶

Principle: Opening = Erosion followed by Dilation. First, erosion removes small noise points, then dilation restores the original size of the objects.

Effect: Removes small noise points while preserving the overall shape of the objects (commonly used for noise reduction).

4. Closing¶

Principle: Closing = Dilation followed by Erosion. First, dilation fills small dark holes, then erosion removes “burrs” from the edges of the expanded objects.

Effect: Fills small dark holes in the image and makes object edges smoother (commonly used to repair broken objects).

Structuring Element (Kernel)¶

The structuring element is a small matrix that defines the “shape” and “size” of the morphological operation. OpenCV creates structuring elements using cv2.getStructuringElement(), with common shapes including:
- cv2.MORPH_RECT: Rectangle
- cv2.MORPH_ELLIPSE: Ellipse
- cv2.MORPH_CROSS: Cross

Code Implementation: From Image Reading to Morphological Operations¶

Step 1: Import Libraries and Prepare the Image¶

First, import the cv2 (OpenCV) and numpy libraries, and read a binary image with noise (if no existing image is available, use cv2.imread() to load a local image).

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

# Read the image (grayscale mode)
img = cv2.imread('noisy_image.png', 0)  # 0 indicates grayscale mode; replace 'noisy_image.png' with the actual image path

# Binarization: Convert the image to a black-and-white binary image (pixels > 127 become 255, others 0)
_, binary = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)

Step 2: Define the Structuring Element¶

Create a 3×3 rectangular structuring element (adjust size and shape as needed):

# Create a 3x3 rectangular structuring element
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))

Step 3: Perform Morphological Operations¶

Apply erosion, dilation, opening, and closing operations, then display the results:

# 1. Erosion operation
erosion = cv2.erode(binary, kernel, iterations=1)

# 2. Dilation operation
dilation = cv2.dilate(binary, kernel, iterations=1)

# 3. Opening (Erosion followed by Dilation)
opening = cv2.morphologyEx(binary, cv2.MORPH_OPEN, kernel)

# 4. Closing (Dilation followed by Erosion)
closing = cv2.morphologyEx(binary, cv2.MORPH_CLOSE, kernel)

Step 4: Display Results¶

Use matplotlib to compare the original image with the processed results:

# Display results
plt.figure(figsize=(12, 8))
plt.subplot(2, 2, 1), plt.imshow(binary, 'gray'), plt.title('Original Image')
plt.subplot(2, 2, 2), plt.imshow(erosion, 'gray'), plt.title('Erosion Result')
plt.subplot(2, 2, 3), plt.imshow(opening, 'gray'), plt.title('Opening Result')
plt.subplot(2, 2, 4), plt.imshow(closing, 'gray'), plt.title('Closing Result')
plt.tight_layout()
plt.show()

Effect Comparison and Explanation¶

Original Image: A binary image with small noise (assumed to contain white noise and a small object).
Erosion Result: Noise is partially removed, but the small object edges shrink and may become smaller.
Opening Result: Noise is completely removed, and the shape of the small object remains largely unchanged (noise removed by erosion, then size restored by dilation).
Closing Result: Small dark holes in the image are filled, and object edges become smoother (holes filled by dilation, then edges optimized by erosion).

Advanced Operations (Optional)¶

Morphological Gradient: Dilation result - Erosion result, used to extract object edges (e.g., cv2.morphologyEx(binary, cv2.MORPH_GRADIENT, kernel)).
Top Hat: Original image - Opening result, used to extract small noise (e.g., cv2.MORPH_TOPHAT).
Black Hat: Closing result - Original image, used to extract small dark holes (e.g., cv2.MORPH_BLACKHAT).

Summary¶

Morphological operations are a foundational tool in image processing. Through basic operations like erosion and dilation, they enable noise removal, object connection, and edge extraction. Beginners can start with opening and closing, adjusting structuring element sizes to gradually master morphological applications in different scenarios.

Practice Suggestion: Try replacing the image path, adjust the structuring element size (e.g., 5×5), and observe the result changes to understand how parameters affect outcomes.