Re: An improved method for scanning documents

["Shawn B." <leabre@html.com>]

> You should've been using the checksums provided next to each listing.  The
> earlier (up to early 1986?) Key Perfect codes would narrow errors down to
> within some group of lines IIRC, but the later codes worked on a
> line-by-line basis and would let you spot errors fairly quickly.  For 
> those
> later codes (don't recall offhand what they called it), they even included 
> a
> program to generate the codes as you typed in the program.  I modified 
> this
> to dump codes for a program already in memory; it wasn't a particularly 
> big
> mod, but it was useful for a high-school student with no spending money. 
> :-)

Easier said than done.  The OCR isn't done on an Apple.  It is done on a 
Windows box.  OmniPage allows you to compare side by side the OCR'd text 
against the actuall text (scanned).  A quick look usually helps with most. 
But, when source lines break a line on the original document, nothing 
automated can help you.  You have to piece the lines together.  Sometimes 
quoted spaces would actually OCR as TAB characters and so on.

In all, the purpose of the OCR is to prevent typing in every line by hand. 
Instead, you have all the text, you just need to verify its accuracy.  Much 
much easier to do.  To get the checksums you speak of to work, I'd have to 
type them in by hand.  Even if I just EXEC them and look at the resulting 
checksum, I still have to correct the lines where 1 (ones) look like "l" 
(small L) and so on.

However, the earlier programs I started working on didn't have checksums 
anyway.


Thanks,
Shawn