Product Reviews

OCR (optical character recognition) is a fairly mature technology. However, it hasn't made any revolutionary leaps in improvement in terms of recognition accuracy in recent years. Instead, the main focus of attention by OCR developers has been on the ability to retain the layout and formatting of an original document accurately.

ScanSoft has introduced some novel features to its OCR package OmniPage Pro 11, including the ability to convert a PDF file into a conventional file that can be edited or exported to other applications. You also have the option of exporting a scanned document as a PDF document. As it's now harder than ever to differentiate the effectiveness of one OCR package over another, ScanSoft's PDF trick is inspired.

We usually expect a new release of OmniPage Pro roughly every 12 months. Version 11, however, arrived 18 months after its predecessor. The delay can be partly put down to the acquisition of Caere, OmniPage's parent company, by ScanSoft at the beginning of last year. ScanSoft now has an embarrassment of riches when it comes to OCR technology. It has software development teams in both the US and Hungary, and it has inherited Caere's own technology plus the OCR engine Caere previously acquired from Calera.

Improvements listed by ScanSoft for OmniPage Pro 11 include a smarter proofing process, improved formatting and accuracy, a despeckle module for dealing with poor-quality original document images, and table detection even when the tabulated data isn't bounded by a grid. You can now also opt to listen to text through multilingual text to speech, OCR directly into Microsoft Word and defer an OCR session using a built-in scheduler.

Previously, OmniPage Pro 10 (see Reviews, issue 64, p196) competed with ScanSoft's formidable and cheaper TextBridge Pro (version 9 reviewed issue 58, p201). Now that they're stablemates, ScanSoft is pitching TextBridge Pro as a budget-priced offering, while

ADVERTISEMENT

OmniPage Pro is clearly aimed at businesses and high-end corporate customers.

As befits the high-end package, OmniPage has a slight edge over TextBridge when it comes to recognition accuracy overall. However, TextBridge did a noticeably better job at deciphering fax-resolution documents. Again, OmniPage can claim a slight lead over TextBridge when rated on layout retention. On the other hand, TextBridge is around 50 per cent quicker than OmniPage at recognising a page - a point to consider if you're planning multipage documents. But to be honest, on pure OCR terms, there's no killer advantage inherent in either one over the other.

Maybe this is what ScanSoft knew all along. When you consider that TextBridge Pro Millennium is bundled in Pagis Pro Millennium - a suite of scanner and file-management tools that retails for around £50 - justifying the undiscounted £370 price tag for OmniPage becomes difficult. But the ace up ScanSoft's sleeve is its PDF capability.

PDF is a great way of distributing information if all you want to do is search it and read it. Reusing that information isn't so easy, though. If you've ever tried to cut and paste text from PDF documents you'll know what I mean. OmniPage Pro 11 can disentangle both the pictures and text from the constraints of a PDF document and export the data to the document file format of your choice. You can also export documents that have resulted from the OCR process as PDFs.

While a PDF original is a formatted document containing text and, maybe, graphics, don't expect OmniPage Pro 11 to make a perfect job of retaining the layout of the PDF original. We found that with some documents, the fonts, font sizes and colours were often incorrectly translated. OmniPage will also try to recognise text in graphics, often with amusing results, though there's an option to turn this off. If all you require is the extraction of the raw text out of PDF documents, it's fine - just don't expect anything more ambitious.

More impressive is the PDF export feature. We were able to create some good PDF documents from scanned documents, though it's a shame you can't use this feature to create PDF documents from existing text documents.

In the end, we were slightly disappointed by OmniPage Pro 11. It still can't deal with fax-resolution input very well and the novel PDF capabilities delivered less than we hoped for. But full marks to ScanSoft for adding a new dimension to OCR and let's hope the PDF functionality can be refined by the time version 12 arrives.