Algorithm for OCR and Kanji Symbols

Question

I have a friend that is starting up a new project. He wants to be able to use some sort of OCR in order to detect and translate Kanji symbols into other languages. He has hit a bit of a brick wall in finding available algorithms in order to do so, since these symbols are a bit more complex than the English characters that we're used to.

We suggested he start looking into 2D convolution and Fourier transforms to start the pattern recognition process, but he is looking for a good starting point.

Unfortunately my knowledge of OCR is extremely limited, so any suggestions that I can pass along will probably be most helpful!

Answer 1

Have a look at nhocr.

(Also, there is tesseract, but I'm not sure if they actually support CJK.)

There are quite a few questions with information about OCR on SO, for instance, try this search.