Prism

Prism is an efficient OCR written in C.

Convert scanned documents and images into editable text files.

Get started

Prerequisites

• GNU Make

sudo apt update
sudo apt install make

• GTK

sudo apt install libgtk-3-dev

Installation

1. Clone the repository.

git clone [email protected]:prismocr/ocr.git

2. Compile the project.

cd ocr
make

Usage

./build/release/ocr [options]

Possible options :

• gui
• ocr [network_path] [image_path]
• sharpen [image_path]
• blur [image_path]
• rotate [image_path] [angle]
• skew_detect [image_path]
• auto_rotate [image_path]
• edge_detect [image_path]
• contrast [image_path] [delta]
• invert [image_path]
• network [xor/add]
• segmentation [image_path]
• network_load [network_path]

Development

Tests

You have to have Unity at the root of the repository:

git clone https://github.com/ThrowTheSwitch/Unity.git

And then:

make test

Auto-format before commit

Add the following code to .git/hooks/pre-commit (create the file if not already existing).

#!/bin/sh

STAGED_FILES=$(git diff --name-only --cached)
make format
git add ${STAGED_FILES}

And make it executable chmod +x .git/hooks/pre-commit

Valgrind script

To compile and run the project there is a script available at ./scripts/valgrind.sh

./scripts/valgrind.sh [target] [program arguments]

For example to run the xor demo in debug:

./scripts/valgrind.sh debug network xor

Authors

Flowtter
aureleoules
TrAyZeN
Obeeron

License

MIT