How can I recognize Japanese characters using OpenCV, C ++?

I have a program that extracts Japanese characters from Manga pages and I want to know what characters they use for character recognition.

I was playing around with the Matching pattern and trying to implement KNearest from scratch

How do I recognize the extracted characters?

I cannot find a good C ++ example about KNearest. I would also like to avoid Tesseract if possible.

I tried pattern matching on the character recognition image and it was detected correctly, but I need it to detect the extracted character. He can't figure it out exactly by playing around with the code http://docs.opencv.org/doc/tutorials/imgproc/histograms/template_matching/template_matching.html

An example of character detection:

Example of extracted symbols:

Example template or training data: