Fine Reader trying to hard would go a long way in explaining the one time
in my recent memory that I had better success with one of Openbook's other
recognition engines. The book was an old worn dirty paperback. As I
recall, the plot wasn't even all that inspiring, and probably didn't
justify the time I put into it. I think it was called the "Five Fingers."
Dave
Yes, fine engine is almost always the best. I have heard that scan soft does better with scanning fractions, but I have no proof of that. Really both scanning engines work quite well, but scan soft tends to do a couple annoying things. I see more punctuation problems with scan soft, such as quotes separated from the sentence with a space, and it doesn't know about the existence of em-dashes, so it just puts in the normal short dash. It also seems less capable of determining whether spaces or line breaks exist, so sometimes you find spaces where they shouldn't be and no blank lines where they should exist.
Fine engines major defect is trying too hard, which you would think was good, but isn't always. :-) It tends to create more junk characters, because if it could possibly imagine something as a character it will put that character in. That results in big globs of junk char mess where pictures are and junk chars at the beginnings and ends of lines where it sees a shadow that is the books binding. The typical binding junk chars include j, i, |, and ; typically attached to a tab character. So, fine engine gets things right, but needs a little junk elimination, whereas scansoft makes some mistakes, but doesn't require exessive deletions.
The RTK engine that Julie refered to no longer is installed by default with new versions of Kurzweil. Good riddens since the only thing that scanning engine had going for it was how fast it recognized. Quick and sloppy, that was its way. :-) Now that computers are faster there really isn't great demand for that.
Sarah Van Oosterwijck Assistive Technology Instructor http://home.earthlink.net/~netentity
