Re: Handwriting Recognition and Future

Holger Freyther wrote:
> Hi Brad, all,
>
> I would like to know the status of your handwriting changes and if they're all
> in CVS. Also you said you would like to work on an improved version of this.
> Do you've any updates on that?

The Graffiti stuff is in and working (I use it daily). As for the rest of the recognition stuff,
not yet, however given that the patent on the 3x3 block recogniser for Graffiti has been invalidated
in the last week or so, I was looking to get going on implementing something along those lines.
Unfortunately life has been a little hectic for the last 3 months and is only just looking like it
will settle down a bit and let me get some coding done.

If someone wants to jump in with ideas then I'm all ears, but I have looked at a number of the
algorithms out there for natural handwriting recognition and found them all pretty cpu intensive.
Plus my handwriting actually is very, very bad, whereas my Graffiti input is passable so for me
Graffiti is probably the direction I'd move in. (I used a Palm Pilot Pro for years and had great
results)

Having said that, I believe that perhaps a 4x4 matrix recogniser could do a better job than the
current Opie recogniser, particularly when you consider a lot of standard Opie characters are
multi-stroke. As an excersize I'm trying to tune the current recogniser to be more accurate but it
has a couple of fundamental flaws.

> On a sidenote what is the Agenda/Status on Opie PIM?

ATM I'm modifying Stefans Pimconverter to try and make it work as a sync agent so I can use
Multisync to sync my PIM databases.

I have been a bit behind for a while (I had not even updated CVS and reloaded my iPAQ for a whole
month!)

Regards,
Brad
Received on Sun Jun 6 01:28:30 2004