Good night:)
I am currently playing with the DevIL library, which allows me to load an image and check RGB values ββper pixel. Just like in a personal training project, I am trying to write a very basic OCR system for several images that I made in Photoshop.
I successfully remove all distortions in the image, and I have text and numbers. I'm not currently looking for an advanced neural network that learns at the input. I want to start quite easily and so I decided to identify the individual characters and count the pixels in these characters.
I have two problems:
- Identification of individual characters.
- Most importantly: I need an algorithm for counting connected pixels (of the same color) without counting the pixels that I previously counted. I have no math background, so this is the biggest problem for me.
Any help on this is appreciated, thanks.
edit:
I marked this question as C ++, because that is what I am currently using. However, pseudo-code or easy-to-read code from another language is also beautiful.
source
share