PDF Text Extraction with Google Vision API

Overview

This project provides a solution for extracting text from PDF files using the Google Vision API. The code converts each page of a PDF into an image, preprocesses the image to enhance OCR accuracy, and then uses the Google Vision API to detect and extract text. The extracted text is saved to a text file for each PDF.

Features

• PDF to Image Conversion: Converts PDF pages to images using the pdf2image library.
• Text Detection: Utilizes the Google Vision API to perform OCR and detect text in the images.
• Image Preprocessing: Enhances OCR accuracy by converting images to grayscale.
• Parallel Processing: Uses ThreadPoolExecutor to process pages concurrently, improving performance.
• Output Formatting: Saves the extracted text in a structured format, with each page's content clearly separated.

Advantages

• High Accuracy: Utilizes Google Vision API, which is known for its advanced machine learning models and high-quality text extraction.
• Efficient Processing: Concurrent processing of PDF pages speeds up the extraction process, especially for large PDFs.
• Customizable: Easily adaptable to different PDFs and output requirements.
• Enhanced OCR: Includes preprocessing steps like grayscale conversion to improve OCR results.

Requirements

• Python 3.x
• Required Python packages:
  • google-cloud-vision
  • pdf2image
  • Pillow
  • numpy
  • opencv-python
  • concurrent.futures

Setup

1. Install Dependencies:

   Install the required Python packages using pip:

   pip install google-cloud-vision pdf2image Pillow numpy opencv-python

2. Google Vision API Credentials:

• Ensure you have a Google Cloud project with the Vision API enabled.

• Download your credentials JSON file.

• Set the path to your credentials JSON file in the environment variable GOOGLE_APPLICATION_CREDENTIALS:

  export GOOGLE_APPLICATION_CREDENTIALS="/path/to/your/credentials.json"
  

3. Usage: Place your PDF files in the pdfs folder. Run the script:

   python script_name.py

Code Explanation

• detect_text_from_image(image_data): Detects text from an image using the Google Vision API.
• preprocess_image(image): Converts the image to grayscale to improve OCR accuracy.
• process_page(page, page_number): Converts a PDF page to an image, preprocesses it, performs OCR, and returns the extracted text.
• process_pdf(pdf_path, output_folder): Processes each page of a PDF and writes the extracted text to a file.
• process_all_pdfs(input_folder, output_folder): Iterates over all PDFs in the input folder and processes each one.