Abstract:

A project was recently added on "garage" [1] to add the CJK (Chinese, Japanese,
Korean) support on the Maemo platform. With about 1.5 billion people using CJK
in the world, this allows the Maemo platform to target a much broader
population.

However, a feature which is extremely common in the PDA with a touchscreen is
still lacking: handwriting recognition. Handwriting recognition is useful when
people want to input a character that they cannot read and thus cannot input
with the standard input method. This of course often happens to foreigners
learning CJK but also to native speakers when it comes to unusual characters.
That is why handwriting recognition makes so much sense for Japanese/Chinese
PDA. Other possible applications of handwriting recognition include dictionaries
and character (kanji) learning games.

This Google Summer of Code project aims to add support for Japanese/Chinese
handwriting recognition using tomoe. [2] Focus will be on performances and
smooth integration with the current input system and latin character
recognition.

[1] https://garage.maemo.org/
[2] http://tomoe.sourceforge.jp

Project details:

Tomoe stands for Tegaki Online MOji recognition Engine and is free software. It
currently supports Japanese only but Chinese is planned. Tomoe is hopefully
divided into several separate modules:

- tomoe: the actual character recognition engine
- libtomoe-gtk: the drawing widget
- scim-tomoe: support for tomoe in scim
- uim-tomoe: support for tomoe in uim

As the core engine, tomoe will probably not have to be modified. For smoother
integration in the maemo platform, libtomoe-gtk will probably not be used "as
is". Some features such as search by reading or number of strokes may have to be
removed because they require a large dictionary called Kanjidic2. However,
libtomoe-gtk will be a good example in order to understand how to use tomoe.

Users should be able to switch the handwriting recognition between latin and
Japanese/Chinese at runtime. Likewise, switch between the virtual keyboard mode
and the handwriting mode should be easy.

When this is done, if time is remaining, time could be used to:

- Add support for recognition of more characters in tomoe. This will require to
improve the tomoe data set. tomoe itself will not have to be modified.

- Improve the current CJK input support (via uim). Possible things to be
improved include: IM switch at run-time, full-screen keyboard and better support
for French accents.

Benefits:

- Support for a very common functionality among Japanese/Chinese PDA
- Maemo targetting a much broader population

About me:

My name is Mathieu Blondel and I am a 23 years old student living in Lille,
France. I am currently in 4th year out of 5 at Telecom Lille1 (French
engineering school, Master degree). I have a sound knowledge of C, C++, Java and
Ruby.

I have been learning Japanese for more than three years and Chinese for a year
and a half. I would like to use my Nokia N800 (thanks to the Nokia discount
program!) as a dictionary and to learn characters. That is why I have a great
interest in the CJK input method and handwriting recognition for the Maemo
platform.

For more details, you can visit my web site www.mblondel.org (resume, pictures,
web log) or contact me at (email removed)

Availability:

Mid-June to end of August is summer holidays in France. So I intend to work
full-time on this project. This summer of code will be my summer job. Before
that, I intend to work on this project on my free time.