I am looking for a OCR solution that is capable of taking an image of a single English alphanumeric character (A-Z, 0-9) and producing the correct alphanumeric result.
As most free OCRs are capable of doing that, there's (of course) a twist!
I'd like the OCR to be capable of recognizing the character regardless of how the character is rotated. Recognition regardless of scale would be a large positive, but not a firm requirement. The ability to recognize the character in the presence of noise is also important. (Minor stray marks or noise in the image should not throw off the OCR significantly.)
I do not care *how* the OCR is implemented, only in results. I don't care if you use OpenCV/feature detection/etc, neural networks, Haar, etc, etc... that's up to you.
-- OCR for alphanumeric character
-- scale-invariant (optional, but recommended)
-- noise tolerant
-- Linux preferred over windows if it is platform-dependent. Either will work.
-- Provide a working example and source code.
Feel free to ask questions if you require further information before bidding.