I'm currently writing an OCR in C++. I'm using a neural network and it works pretty well with simple letters like this.
But now I would like to go further and extract letters out of an image, like this one.
I'm using OpenCV 3 and I have no clue of how I could do this, I searched a lot on internet but I didn't find anything that could help me.
Right know I'm able to crop out most of the words by using findContours, but it seems impossible to get each letter separately.
Any help/idea would be really appreciated !