I used to work in the Division of Planning, in the Kentucky Transportation Cabinet. Also in that Division was the Cartographic Section. About twenty-five years ago they tried to use early OCR when making updated maps and quickly gave up on it. They tried again, several times over the years, and eventually found something acceptable for recognizing print and deciphering it. The final system used files which contained both image elements and characters. You could go in, select the name of a road, change it, check the spelling and save it.
I only used this for a couple of maps I did when they were behind about five years ago. I don't remember the software. I recall it was specifically for mixed graphics functions, such as making maps. They had only been using it for (IIRC) a couple of years before that. It worked quite well. It was also expensive, both in itself and in hardware requirements.
Something like this would probably work for making books, but if you have a good quality image scan why not just use that, like you mention? Having extensive experience with large bureaucracies (Government employee, remember?) I think someone, in upper management, is convinced that OCR is far more futuristic and advanced than images. Especially since they found a real bargain on an OCR package. No sense wasting money for the same thing, right? Besides, text files - even with formatting code - are smaller than high-quality image files, so they save space. That's good, right?
I have a reprint book made from microfilm. There are some minor flaws, but the few typos are from the original.
no subject
I used to work in the Division of Planning, in the Kentucky Transportation Cabinet. Also in that Division was the Cartographic Section. About twenty-five years ago they tried to use early OCR when making updated maps and quickly gave up on it. They tried again, several times over the years, and eventually found something acceptable for recognizing print and deciphering it. The final system used files which contained both image elements and characters. You could go in, select the name of a road, change it, check the spelling and save it.
I only used this for a couple of maps I did when they were behind about five years ago. I don't remember the software. I recall it was specifically for mixed graphics functions, such as making maps. They had only been using it for (IIRC) a couple of years before that. It worked quite well. It was also expensive, both in itself and in hardware requirements.
Something like this would probably work for making books, but if you have a good quality image scan why not just use that, like you mention? Having extensive experience with large bureaucracies (Government employee, remember?) I think someone, in upper management, is convinced that OCR is far more futuristic and advanced than images. Especially since they found a real bargain on an OCR package. No sense wasting money for the same thing, right? Besides, text files - even with formatting code - are smaller than high-quality image files, so they save space. That's good, right?
I have a reprint book made from microfilm. There are some minor flaws, but the few typos are from the original.