How can get digits from led secreen with EmguCV?

Question

I am working on a project. Which captures frame with cam and detect characters with using OCR. When i developed this project i benefited from EMGU CV's standart "OCR" and "Capture" examples. Now, i can capture frame but i did'nt get real OCR result yet. Do you suggest an example or method for doing this.

PS my english is not enough for reading so technical article. Because of this reason, if your suggestion contains the code, that will be good for me.

Thank very much for your help, already now. Example, frame sources like below;

Answer 1

I am sorry but I have to refer you to an article: http://www.codeproject.com/Articles/196168/Contour-Analysis-for-Image-Recognition-in-C

But don't read it, just download the demo application with the code. Run the demo and install an LED font on your computer.

If this works for your project, analyze the code and move accordingly.

Answer 2

If your images will always be the same, you could use another OCR engine, this being Tesseract . This however depends on your images, since different images will require the Neural Network different training patterns. So, if you only have two types of images as shown above, you might be able to get away with it with one or two instances of the OCR engine (one for each format). That being said however, things could get difficult once more image patterns are thrown in.

Since EmguCV is a wrapper for OpenCV, you could use the library for some image processing before passing it to the OCR.