Geek Answers Handbook

OCR (Optical Character Recognition) for on-screen text

I am trying to create a piece of software that automates a PC by capturing a screenshot, then OCR (optical character recognition) that looks for a specific button to click (for example). I have a mouse and keyboard control, but now I needed OCR to process the screenshot. I found that Tesseract OCR does not seem to work very well with text on the screen. The text is too small or that some text seems to be connected, for example, K and X. How do I do this?

p / s: this is for an automated test program.

+4
source share
2 answers

Perhaps check out this question about improving your image to OCR. Otherwise, this question is very similar to "OCR for.NET" .

If you feel really brave, you can always crack a simple approach based on a perceptron or neural network :-)

0
source

I'm not sure if this really suits you, but some of the best OCR I've seen in automation are made by Tevron CitraTest . It has a font library, and if there is no font, they will create a new one based on your ideas. The beneficial factors with this tool will be the costs and common problems associated with changing the screen resolution.

0
source

More articles:


All Articles