CBIR System in MATLAB

This project implements a Content-Based Image Retrieval (CBIR) system in MATLAB.
The system extracts features from images, compares them to a query image, and returns the top-3 most similar images from the dataset.

---

Features

• Extracts 9 features (intensity, morphological, and texture-based) from each image:
  • Mean intensity
  • Standard deviation of intensity
  • Minimum intensity
  • Maximum intensity
  • Area (sum of pixel values)
  • Entropy (texture feature)
  • Energy (texture feature)
  • Centroid X, Centroid Y (from morphological analysis)
• Divides each image into 16 equal blocks and extracts features per block.
• Computes similarity between query and dataset images using Euclidean distance via pdist2.
• Includes a separate MATLAB script to resize dataset and query images before feature extraction.
• Returns and displays the Top-3 most similar images to the query.

---

Dataset

This project uses the Corel Image Dataset available on Kaggle:
Download dataset

---

Example Output

• Input query: query.jpg
• Returned results:
  • Rank 1: Most similar image
  • Rank 2: Second most similar
  • Rank 3: Third most similar

---

Author

Shady Nikooei
Digital Image Processing