Using Tesseract to scan for object in image and read characters

[Emerson Walsh]

Hi,

I have been trying to use tesseract-ocr on my Raspberry-Pi to read images I have saved. I have had decent success so far but hope to take this much further. My main goal is to have it scan an image I give it, search for a specific object in the picture (a square with a 4 digit number inside), and once it finds it, use OCR to send back the numerical data (the number inside the object). As of now, I have really only been working with different ways of manipulating the image I give it, so it is easier for tesseract to read.

Do you have any advice for me moving forward (especially with a way to scan an image for the specific object)? Is this possible? Any specific software you know that could help me?

I really appreciate any feedback at all.

Thanks,

Emerson

[Emerson Walsh]

Anybody willing to help?

[Paul]

This paper might also be relevant to you: http://www.imlab.jp/cbdar2007/proceedings/papers/O1-1.pdf
But it might be to much for the RaspberryPi.

You could binarize the image with this method, then do OCR on it.