Recognise numbers and replace them with text

[tomerb15]

Is there a way to use tesseract to find the numbers in an Image and get the x,y of the numbers in the image?

I have numbers which acts as placeholders in the image (must be numbers) and they can be in various rotations within the image. (sometimes the numbers are upside down/sideways).

I need to replace the numbers with some text

[reecefenwick]

Tesseract can output hOCR file which is basically a html representation of the OCR results.

Gives you words and their coordinates, confidence and much more.

I'm unsure node-tesseract allows for this output, but you should be able to get it easily by running tesseract on the cli.

http://stackoverflow.com/q/8268928

You can then use something like this https://github.com/gierschv/node-hocr to parse the html into nice objects.

I use tesseract with my Java app and follow a similar process, get hocr, parse it, analyse coordinates of words etc

If I can find some time I'll see if I can getting something working in node.

[tomerb15]

Thank you very much for your kind help :)