Shane <shane-keyword-speakup,aca783&cm,nu> envisioned:
> Another lingering use for the Windows machine
<laughs> At least the [CENSORED] thing helped me get Linux
running. That, and help from folks here and the cvale LUG.
Unfortunately, it seems I've already been hacked. Looks like I'll have
to do a full, scratch Slackware reinstall. Found some hidden files in
my home directory that aren't even root accessible. At least I can
expect Windows to be hacked, so don't care what really happens to that
particular OS. It's made to be reinstalled from scratch anyway.
> unfortunately. Only software I know can do this is
> Kurzweil and Nuance OmniPage. However, with Google's OCR
> project in developement I wouldn't be at all surprised if
> something like this becomes possible under Linux at some
> point.
Yeah, I'm thinking that the OCR algorythms would need to include
font recognition and size, and location information in the scanned
image. Font size and location data should already be variables within
the OCR code, and possibly even font type. I think the hard part would
be in deciding whether to make the program a stand-alone project, or to
modify the current OCR routines to communicate more info to stdout.
Maybe in HTML format, so you could use a browser to fill in the
blanks. That, or maybe postscript, as long as it's not proprietary,
like Adobe's PDF scheme. Oh well. Something to while away the hours
in the future, I guess.
Michael
