During the last part of this year I have gone back and played with some technologies that I experienced years ago.
First, I went back and had a look at voice dictation software which I found has matured into a very usable product. And then the other day, not feeling like typing a photocopied document into my computer, I wondered how optical character recognition (OCR) software was coming along.
OCR software will translate non-editable digital images of text into editable text files and was something that I have played with on a number of occasions in the past. It wasn’t that good then but, in the light of my experience with the improved dictation software, I decided I would give optical character recognition another go.
I had a look on the Internet and found that OmniPage seems to be an industry-standard at the moment and I had a hunt around to see if I could find a trial version of the program. OmniPage is owned by Nuance and can be found on the Internet at www.nuance.com but I couldn’t seem to find a trial version of the current program, which is version 16.
I did find the Nuance Australian site is still offering the previous one, version 15, of OmniPage Professional ,on a 15-day free trial basis. I soon got the program installed and found that the interface is really very simple, although it might look a bit complex at first glance.
The first thing you do when you want to use the software, is to go to the load files facility, which allows you to load pictures of the text you want the program to recognise. These are digital pictures of text created with a scanner or digital camera, and I tried both, having taken pictures of some pages from a book ,and scanned in more pages using my desktop scanner.
Once the files are loaded, you go to the recognition section where there is a fully automatic mode in which the software automatically tries to gauge which parts of the image are the text you want it to recognise. It’s probably easiest to try the automatic setting first but you can manually control the result by telling OmniPage where the text on the page is, which bits are tables or pictures, and which parts it should leave out.
Once it has completed processing the image files, it goes into proofing mode where it highlights suspected errors in the text and gives you a chance to correct them by referring to the original image file. You can save your text as a plain text file, a Word file complete with the pictures and formatting from the original, or in any of a number of other different formats.
I have used OmniPage 15 Professional quite extensively and I’m amazed at the quality of its recognition, providing you give it a decent original to work with. When I used my desktop scanner, or my digital camera to capture the image of pages out of books, or whatever, recognition was almost perfect.
The results were not so good at first, when the images I fed into OmniPage were poor because I took them with my digital camera on a low resolution setting’. It did make an attempt at recognising the text but, once I re-took the pictures at high resolution, the output accuracy shot up to nearly 100%.
As I said before, I was very impressed with the quality of the program and am going to track down a South African supplier as soon as I can. The OmniPage 16 Standard version is available for $149 and, looking at the Nuance.com site, it looks very much as if it has most of the features except for some abilities to process files automatically.
Anyway, it was a revelation to me because it is going to save me a lot of time when it comes to digitising material for my Facts About Durban (fad.co.za) history website.
Why not leave a comment by clicking the link below?