I’m currently making a system that will need to be able to read a landline’s screen and process the text on it.
Right now, I’m using edge detection to find the actual screen which seems to work fine. I’ve also applied erode and dilate to try and connect the numbers. The threshold image doesn’t look that bad but when I hand it over to the OCR it’s not able to recognize almost any of the text.
I’ve tried applying blurs and I also gave it a sharpened version of the image but it still struggled to read it.
I think this might be more of an OCR issue rather than needing to process the image more but I’m not sure.
Does anyone have any tips on where I can go from here?
Thank you.
