Fun with OpenCV and Python

The main goal of this script is to :

1. Open a photo of a Suduko Grid
2. Filter and correct it (perspective correction, noise suppression...)
3. Identify the grid
4. Read the content of the cases
5. Solve the Sudoku
6. Display the solution

The libraries used are OpenCV2 and Numpy

Despite my efforts, the OCR doesn't seem to be very accurate (50-60%), which is not enough for our application...