OmniPage Pro 8

Caere's simple selling point for OmniPage Pro is that it's the most accurate OCR package on the market - bar none. Version 8 features a new OCR engine, impressively titled Predictive Optical Word Recognition Plus. This checks the different permutations for the characters that make up a word and decides on the most likely letter combinations. In other words, it doesn't simply identify letters by their outline or density, but instead works out what the most likely following letters are and attempts to recognise those as well. This results in quicker, more accurate OCR. I found it to be around 30 per cent faster than TextBridge Pro 98 (reviewed p208).

OmniPage 8 is able to handle text skewed up to ten per cent. In fact, it's better at reading text at this angle than I am. The results become fairly variable when you get near the ten per cent limit though. Just like TextBridge, it also handles white text on black and dropped capitals. The most revolutionary feature, however, is its ability to deal with text produced in any one of 13 European languages. An example of where this is useful is that, in English, the letter Q is almost always followed by U, whereas this isn't the case in several European languages.

OmniPage spoils Word users in that it shares Word's spelling dictionary to save you the hassle of redefining custom words for both, and you can access OmniPage from the Word File menu. OmniPage will run within 17 word processors and add the recognised text via the clipboard as an RTF file. This process is similar to that with TextBridge, except it's simpler and quicker. The only evidence you have that anything is happening is a couple of progress bars, which is neater.

OmniPage can also now output to HTML, although the results are no better than TextBridge: you won't be able to manage without significant further editing. One area where TextBridge does have an advantage over OmniPage is in its ability to output to Adobe Acrobat. If that's tempting, then TextBridge is the better option.

On the productivity side, OmniPage now supports Multithreading, which means that it allows the user to edit one page while scanning or recognising a number of others. It has a more robust and comprehensive batch processing feature than TextBridge and allows you to specify a time for this to take place - overnight perhaps.

At the heart of any OCR package is the accuracy of its character recognition, and along with the OCR engine, OmniPage employs a number of technologies to improve this. These include AnyFont technology, which accurately maps OCRed fonts to those on the resident PC, including emboldening, italics and underline. This technology, like the others, is transparent to the end user but gives excellent results, particularly with challenging documents.

Also, 3D OCR gives OmniPage a major advantage by handling greyscale images, making it much better at recognising text on coloured backgrounds as well as converting any images on the page more satisfactorily. This feature is likely to be top of Xerox's wish list for the next release of TextBridge.

OmniPage also includes AnyFax Technology. As well as being able to recognise faxed messages (although low-resolution and thermal paper faxes still yield variable results), AnyFax is able to use the CCITT codes included with faxes received on a PC to make sure that the data is complete.

Despite all this technology behind it, the process of scanning and recognising documents is very simple. Using OmniPage as a standalone application reveals a toolbar that's similar to, but more comprehensive than, TextBridge's. Clicking the AutoOCR button starts the OCR Wizard which is, again, more comprehensive than its rival's. The Wizard allows the user to specify settings such as whether the image should come from scanner, stored bitmap or Microsoft Exchange, along with page layout options.