Ocr Sollution

[Memoras]

heyho

ive been writing on a bot for a flash based browser game for some time now.

My problem is that i dont got a really good sollution to read out the text displayed.

I tried jocr/gocr but without satisfying results

I got the most accurate resulst with ABBYY Fine Reader Screenshot reader (id say 99% correct results) but id really prefer to have a freeware/open source sollution for this

You got an advice for me how to improve the results of gocr oder any other program?

Sorry if this has already been asked but i didnt found an answer

cheers Memo

[FreeFry]

Is it a static text(ie. it doesn't change)? If so you could perhaps use PixelChecksum to conclude if the text you're looking for is there. But if it's something such as a score(or if the background changes, for example), you can't use PixelChecksum.

[Memoras]

I dont think thats possible

The main goal of it is to identify an item by its name, which consists of basic items name + prefix for the stats

If i see this right i would have to get the pixelchecksum for every item there is, which would be way too much.

But thanks for this idea i think it helps me on another point of the bot

[FreeFry]

If you're just after the item name, you could just do a pixel checksum for the item name(not the following prefix, etc.), and then compare (with another PixelChecksum) the area where the item name is being display. Hopefully you can use it that to identify the item you're after. As long as the background around the text, and the location doesn't change(colorwise that is), it should work fine.