Document Scanner

A Python-based document scanner application that automatically detects document edges, applies perspective transformation, and enhances the document for better readability.

Project Overview

This document scanner uses computer vision techniques to:

• Detect document edges in an image
• Apply perspective transformation to obtain a top-down view
• Enhance document contrast and readability
• Process and save the scanned result

Technologies Used

• Python 3.11
• OpenCV (Computer Vision library)
• NumPy (Numerical computing)
• Pillow (Image processing)
• imutils (Image processing utilities)
• scikit-image (Image processing algorithms)

Installation

1. Clone the repository:

   git clone https://github.com/TrishamBP/document-scanner.git
   cd document-scanner

2. Create a virtual environment:

   python -m venv venv

3. Activate the virtual environment:
   • Windows:

     venv\Scripts\activate

   • macOS/Linux:

     source venv/bin/activate

4. Install the required packages:

   pip install -r requirements.txt

Usage

1. Place the document image you want to scan in the project directory or specify the path when running the program.
2. Run the main script:

   python scan.py --image path/to/document.jpg

3. Optional arguments:

   python scan.py --image path/to/document.jpg --output scanned_result.jpg

4. The program will:
   • Detect document edges
   • Apply perspective transformation
   • Enhance the document
   • Save the processed image

Implementation Details

The document scanner follows these key steps:

1. Edge Detection: Using Canny edge detection to identify document boundaries.
2. Contour Detection: Finding the document contour within the image.
3. Perspective Transform: Warping the image to get a top-down view of the document.
4. Enhancement: Applying adaptive thresholding and other techniques to improve readability.

Project Structure

📂 Document-Scanner
│-- 📂 examples/ (Sample images before and after scanning)
│-- 📜 scan.py (Main script for document scanning)
│-- 📜 requirements.txt (Dependencies)
│-- 📜 README.md (Project documentation)

Examples

Original Image	Scanned Result

License

This project is licensed under the MIT License - see the LICENSE file for details.

Future Improvements

• Add OCR (Optical Character Recognition) functionality
• Implement batch processing for multiple documents
• Create a web interface for easier usage
• Add mobile support through a REST API

Acknowledgements

• Adrian Rosebrock's PyImageSearch tutorials
• OpenCV documentation and community
• All open-source libraries used in this project