theolepage / ocr

Optical Character Recognition software based on a simple NN and written from scratch in C.

OCR

Project

OCR stands for Optical Character Recognition software. As part of a student project, our objective is to make a program capable of extracting text from images. It must be written in the C language, rely on a neural network and be used through a GUI. To learn more, please take a look at the book of specifications.

Usage

Dependencies: SDL2, SDL2_image, GTK+ 3 and Hunspell.

1. Clone this repository with git clone [e-mail]:NoneOfAllOfTheAbove/OCR.git.
2. Compile the project by running the command make in the project folder.
3. Execute the program with ./bin/OCR.

Features

Currently implemented:

• Advanced preprocessing (efficient binarization, noise canceling, contrast enhancement)
• Detect paragraphs, lines, words and characters
• A pretrained neural network to recognize characters
• Simple GUI to load an image and export its extracted text
• Postprocessing step (spell check)

Features we are working on:

• De-skew
• Improve segmentation (export as HTML)
• Retrain the neural network

Contributing

Refer to CONTRIBUTING.md.