Skip to content
/ OCR Public

Repository of an Optical Character Recognition implemented in C with multiple modules : binarization, segmentation, neural network and an interface !

Notifications You must be signed in to change notification settings

Bictole/OCR

Repository files navigation

OCR Profile

The OCR Logo

OCR is an Optical Character Recognition made in C. The goal was to perform a binarization, a segmentation and send the matrices in a neural network to recognize the characters in an image.

We have :

  • Black_White module that is supposed to binarize the images into a black and white only image.
  • Interface module which contain the interface program, made with SDL 2.0
  • Neural_Network module, the machine learning part of the OCR.
  • Reconstruction module, this part comes after the segmentation and the neural network, where our code needs to reconstruct the initial text from the image into a buffer.
  • Segmentation module is the part of the program where the input image is segmented into different paragraphs, lines, words and finally letters and that send the pixels matrices into the neural network.
  • Tools module, contains several uselful tools, especially for the SDL library manipulation.

How It Works

The OCR contains a CLI tool which is used to demonstrate the segmentation part of the project.

Segmented image

Usage

make
./src/main

The trash files produced by the compilation could be removed by :

make clean

or

make properclean

Interface

The interface is made with SDL 2.0, it's a really simple component that need to be refactored. Here is a diagram that demonstrate its usability.

Interface

Costs

The outrageous Costs

About

Repository of an Optical Character Recognition implemented in C with multiple modules : binarization, segmentation, neural network and an interface !

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published