This kanji handwriting recognizer, which is dependent of the correct stroke order and stroke numbers, uses the KanjiVG data as templates. As the first step, a kanji is translated into a code sequence as shown in the graphs.
The code sequence of 京 is "6161B1556".
The recognizer uses levenshtein distance to calculate the difference between 2 code sequences.
Install the levenshtein method, which is a Ruby native extension
$ git clone https://github.com/thebluber/kvg_kanji_recognizer
$ cd levenshtein
$ make installSetup and run server on port 9292
$ bundle install
$ bundle exec rackupTest the API
$ curl -X POST -H "Content-Type: application/json" -d @strokes.json http://localhost:9292/api/v1/scores
=> {"scores":["二","冫","于","土","士","工","干","七","下","丶"]}