Document Scanner

A Python application that processes images of documents, automatically detects their boundaries, applies perspective correction, and converts them to enhanced PDF files.

Features

• Automatic document boundary detection
• Perspective correction
• Text skew correction
• Image enhancement and binarization
• Batch processing of multiple images
• Automatic conversion to PDF

Requirements

opencv-python
numpy
Pillow

Installation

1. Clone this repository or download the source code
2. Install the required dependencies:

   pip install opencv-python numpy Pillow

Usage

1. Create an input folder in the project directory
2. Place your document images (PNG, JPG, JPEG) in the input folder
3. Run the script:

   python app.py

4. Processed PDFs will be saved in the output folder

How it works

The application performs the following steps:

1. Loads and preprocesses the image (grayscale conversion and blur)
2. Detects document boundaries using contour detection
3. Applies perspective transformation to get a top-down view
4. Corrects text skew using the Hough transform
5. Enhances the image using adaptive thresholding
6. Saves the result as a PDF

Error Handling

The script includes error handling for:

• Invalid or unreadable images
• Cases where document boundaries cannot be detected
• File system operations

Notes

• The input images should have the document clearly visible with good contrast from the background
• Supported input formats: PNG, JPG, JPEG
• Output files are saved as PDF with the same name as the input file