Comparison of Connectivity in Image Segmentation

Overview

This project demonstrates the use of connectedness for image segmentation in binary images. It focuses on comparing 4-connectivity and 8-connectivity, widely used methods to identify connected components in an image. By visualizing the results side by side, the project showcases how different connectivity rules affect segmentation outcomes.

Features

• Binary image thresholding to simplify segmentation.
• 4-connectivity:
  • Groups pixels based on horizontal and vertical adjacency.
• 8-connectivity:
  • Groups pixels based on horizontal, vertical, and diagonal adjacency.
• Color-coded visualization for connected components.

Requirements

• Python 3.x
• OpenCV library (cv2)
• NumPy library (numpy)
• Matplotlib library (matplotlib)

Installation

Run the following commands to install the required libraries:

pip install opencv-python
pip install numpy
pip install matplotlib

Usage

1. Place the input image in the project directory.
2. Update the image_path variable in the code with the name of your image file.
3. Run the script to generate the segmentation results and visualizations:

python connectivity_comparison.py

Output

The script produces the following visualizations:

1. Original Image: Displays the grayscale input image.
2. Binary Image: Shows the binarized version of the input image after thresholding.
3. 4-Connectivity Segmentation: Illustrates connected components using 4-connectivity rules.
4. 8-Connectivity Segmentation: Illustrates connected components using 8-connectivity rules.

The connected components are visualized with random colors to highlight distinct regions.

Notes

• Connectivity Difference:
  • 4-connectivity is stricter and may result in smaller connected regions as it ignores diagonal relationships.
  • 8-connectivity is more flexible and often results in larger connected regions due to diagonal inclusion.
• You can experiment with different threshold values to analyze how segmentation changes.

Future Enhancements

• Add support for custom connectivity rules (e.g., extended connectivity beyond 8-connected neighborhoods).
• Apply connected component analysis to multi-channel (color) images for advanced segmentation.