Skip to content

Latest commit

 

History

History
41 lines (33 loc) · 2.24 KB

README.md

File metadata and controls

41 lines (33 loc) · 2.24 KB

Image Segmentation using Lazy Snapping and K-Means Clustering from Scratch

Overview

This Jupyter Notebook demonstrates image segmentation techniques using two popular methods: Lazy Snapping and K-means clustering. The objective is to partition the images into meaningful segments based on pixel intensity similarities and user-defined masks. This notebook showcases how different values of K impact the segmentation quality and compares the results of both algorithms.

Dependencies

  • Jupyter Notebook
  • Python
  • NumPy
  • matplotlib

Features

  • Image Segmentation: Segment images using Lazy Snapping and K-Means Clustering.
  • Comparison of Segmentation Quality: Evaluate the impact of k values (2, 4, 8, 16) on segmentation.
  • Visual Representation: Display the original and segmented images in a grid for easy comparison. image

Implementation

The notebook is structured into several key steps:

1)K-Means Clustering:

  • Implement the K-Means algorithm to partition the image into k clusters based on pixel intensity similarities.
  • Update centroids iteratively until convergence.

2)Lazy Snapping:

  • Extract seed pixels from user-defined foreground and background masks.
  • Compute the likelihood of each pixel belonging to the foreground or background class.
  • Assign pixels to classes based on likelihood values to obtain segmented images.

3)Visualization:

Display the original image and its segmented counterparts obtained through Lazy Snapping and K-Means Clustering for different k values. image

Usage

  • Load the Image: Load the image along with the foreground and background masks.
  • Set Values of k: Specify different values of k to observe the impact on segmentation.

-Segmentation Process:

  • Use the Lazy Snapping method to segment the image based on user-defined masks and seed pixels.
  • Apply K-Means Clustering to partition the image into k clusters.
  • Visualize Results: Display the original and segmented images in a grid for comparison.