Product Reviews

Caere's OmniPage PRO is one of the leading optical character recognition applications, a tool which reduces the amount of typing needed to transcribe documents by recognising words from scanned pages. Although this kind of software doesn't usually cause much excitement, it can make life in an office much easier, so major upgrades shouldn't be dismissed.

Caere has taken two and a half years to develop OmniPage Pro 8.0. The program is based around an entirely new OCR engine, which the company claims makes it both faster and more accurate than previous versions. The interface has also been simplified in order to make the OCR process easy for the user. It now consists of a number of separate windows, including a toolbar with a few pop-up menus and large buttons which provide access to virtually all the main features. Provided the main settings are to the user's liking, the whole process can be set in motion by clicking on the Auto button. This runs OmniPage in hands-free mode, driving the scanner (or reading scans from disk), optimising them, performing the OCR and even saving them in one of a variety of different file formats. With a collection of pre-scanned images or a sheet feeder-equipped scanner, this allows a Mac to process a large number of pages, even at a pre-determined time, without further intervention.

If more control is needed - for instance, to specify particular areas to recognise or ignore, or to identify numeric-only sections - the process can be performed manually by selecting options from pop-up menus then clicking on the corresponding buttons. These are laid out in a logical 'scan, set-up, OCR, save' sequence, so it's easy to customise things and get results with very little practice.

Caere claims accuracy levels of 99%, which sounds pretty impressive, even when you remember this is per character, not per word. This level of success is reliant on using clearly printed documents using standard fonts and type sizes, not well-thumbed pages with annotations and coffee stains.

OmniPage can recognise text as tables and multiple columns as well as simple single-column data. It can also recognise

ADVERTISEMENT

and preserve colour text, and can attempt to preserve the font and style formatting of the original document. It even goes as far as recognising images in the page and storing them as embedded pictures, and the results can be saved in one of a number of graphics- and layout-preserving document formats such as Microsoft Word 5.0 or 6.0, RTF or WordPerfect 3 formats.

One particularly useful improvement is in the way the program handles skewed scans. Caere claims OmniPage can work with scans which are tipped up to 10¡, a distinct improvement on previous versions. It handles these by performing a 'straightness recognition' and adjusting the orientation automatically, although this can be done by hand if you prefer.

The core OCR process is based on character recognition, but the OCR engine uses one or more dictionary files to perform on-the-fly word recognition as well. This makes the results and proof-reading suggestions a little more intelligent, and also provides the basis for the multilingual OCR claims.

OmniPage ships with dictionaries for a number of different languages, including UK English, Danish, French, German, Spanish and two varieties of Portuguese. These can be enabled individually or together, and are used by OmniPage when recognising the text and performing the proof-reading functions. This allows it to handle multiple languages on a page with aplomb, although enabling many different dictionaries will slow OmniPage down. It struggled a little with our German test page, but this was due to the quality of the photocopy rather than any discernible language deficiencies. It handled the language well and would certainly be useful for transcribing multilingual documents of any reasonable visual clarity.

Results are, of course, limited by the quality of the scan, so the scanner settings can affect the OCR results. OmniPage provides some basic scanner driver settings within its own preferences. Not every device listed as 'compatible' can be controlled directly from OmniPage, although in these cases the scanner's own software will usually work with the program. The Visioneer PaperPort Strobe, for instance, can pass scans directly to OmniPage Pro 8.0.

How much you need an OCR package is partially determined by how fast and accurate the software is. Smudged photocopies are always going to be a problem, but OmniPage Pro 8.0 has certainly improved your chances. Caere's ease-of-use claims were borne out by our tests, and the general accuracy and speed was impressive. The program copes with pages which previous versions and competing applications simply choke on. If you need to retype documents on a regular basis and have access to a scanner, then OmniPage Pro 8.0 is likely to be a help.