doc-img-extractor

Tool for extracting images from documents like PDF, DJVU, EPUB, DOC, DOCX using system packages. Includes multithreaded OpenCV figure extraction and OCR verification.

THIS TOOL IS NOT PERFECT. THERE ARE STILL FALSE POSITIVES AND FALSE NEGATIVES. BEWARE.

Features:

• Multi-format support: PDF, DJVU, EPUB, DOC, DOCX.
• Smart Extraction: Renders pages to detect embedded vector figures (OpenCV).
• OCR Verification: Uses Tesseract to distinguish figures from text blocks.
• Multithreaded: Utilizes all CPU cores for faster processing (toggleable).

Required Packages

To compile and run this tool, you need both runtime dependencies (to process files) and development libraries (to compile the code).

Debian, Ubuntu, Linux Mint (Debian-based)

Run the following command to install everything needed:

sudo apt-get update && sudo apt-get install -y \
    build-essential \
    libfltk1.3-dev \
    libopencv-dev \
    tesseract-ocr \
    libtesseract-dev \
    libleptonica-dev \
    djvulibre-bin \
    libreoffice \
    poppler-utils \
    unzip

Fedora (and other RHEL-based distributions like CentOS, AlmaLinux)

sudo dnf install -y \
    gcc-c++ \
    fltk-devel \
    opencv-devel \
    tesseract \
    tesseract-devel \
    leptonica-devel \
    djvulibre \
    libreoffice \
    poppler-utils \
    unzip

Compilation

Compile the program using g++. We use pkg-config to automatically locate the correct library paths and dependencies for FLTK and OpenCV, and explicitly link Tesseract.

g++ -std=c++17 main.cpp -o a \
    `fltk-config --cflags --ldflags` \
    `pkg-config --cflags --libs opencv4` \
    -ltesseract -llept -pthread

Flags explained:

• -std=c++17: Ensures C++17 support for multithreading features.
• fltk-config --cflags --ldflags: Automatically handles FLTK dependencies.
• pkg-config ... opencv4: Automatically handles OpenCV dependencies.
• -ltesseract -llept: Links the OCR libraries.
• -pthread: Ensures proper threading support for the application.

Usage

1. Run the executable:

   ./a

2. Click "Choose" to select your documents.
3. Select an Output Directory.
4. Configure options:
   • Enable OpenCV: Check to render pages and detect figures (slower, more accurate).
   • Use OCR: Check to verify if a region is text or an image (slower).
   • Enable Multithreading: Check to use 100% CPU for rendering.
5. Click Start.