Arabic Font Recognition - README

---

Project Overview

This project implements an Arabic font recognition system by processing images containing Arabic text. The system identifies fonts through the following steps:

1. Noise Removal: Cleans the image to enhance text regions.
2. Segmentation: Breaks the image into text regions, lines, words, and characters.
3. Orientation Detection: Aligns the text for proper processing.
4. Feature Extraction: Extracts features using methods like bag-of-words or SIFT.
5. Font Classification: Uses a trained classifier to predict the font based on extracted features.

---

Features

• Arabic text segmentation into lines, words, and characters.
• Noise removal for better segmentation accuracy.
• Orientation detection and correction.
• Font recognition through feature extraction and classification.

---

Folder Structure

• CharCutter.py: Extracts individual characters from words.
• Classifier.py: Implements the font classification model.
• FeatureExtractor.py: Handles feature extraction methods.
• ImageLoader.py: Loads and visualizes input images.
• LineCutter.py: Segments images into individual lines.
• NoiseRemoval.py: Reduces noise in the image.
• OrientationDetector.py: Corrects the orientation of text.
• Preprocessing.py: Prepares images for further analysis.
• Segmentation.py: Splits images into meaningful segments (e.g., lines).
• SeparationRegion.py: Helps identify regions between characters or words.
• WordCutter.py: Segments lines into individual words.
• main.ipynb: Example notebook showcasing the pipeline.
• renaming.py: Utility script for renaming files.

---

How to Run

1. Install required dependencies:

   pip install -r requirements.txt

2. Place your dataset in the fonts-dataset directory, structured by font types.
3. Run the main script or the notebook to process images and classify fonts:

   python main.py

---

Usage Example

The following example showcases the pipeline's functionality:

from ImageLoader import *
from NoiseRemoval import *
from Segmentation import *
from OrientationDetector import *
from FeatureExtractor import *
from Classifier import *

# Load an image
path = "./fonts-dataset/2/"
image = ImageLoader.loadImage(path, "1.jpeg")

# Process the image
image = NoiseRemoval.applyGaussianBlur(image)
image = Segmentation.segment(image)
image = OrientationDetector.rotate(image)

# Extract features and classify
feature_extractor = FeatureExtractor()
feature_extractor.loadDataset("./fonts-dataset/")
features, labels = feature_extractor.bagOfWords()

classifier = Classifier()
classifier.train(features, labels)
result = classifier.classify(image[0], feature_extractor, method='SIFT')
print(f"Predicted Font: {result}")

---

Future Enhancements

• Improve segmentation accuracy for complex layouts.
• Add support for more robust feature extraction techniques.
• Explore deep learning-based methods for classification.

---

Feel free to contribute or suggest improvements! 😊