Braille Monitor                                                    November 2008

(back) (contents) (next)

Optical Character Recognition and High-Volume Book-Scanning

by Clara Van Gerven

Clara Van Gerven demonstrating a Zoom-TwixFrom the Editor: Clara Van Gerven is an access technology content specialist in the International Braille and Technology Center of the NFB Jernigan Institute. The following article summarizes a presentation she made at the 2008 NFB convention. Here it is:

Few things have become as important in access technology over the years as optical character recognition (OCR). As the primary way to access print, OCR has become an important tool for any blind student or professional, reducing (though by no means eliminating) the need for human readers. The simple function of converting print to text readable by a text-to-speech engine has long since been expanded to include a wide range of editing and formatting choices. This article will describe and compare four of the most commonly used packages in OCR: Freedom Scientific's OpenBook 8, Kurzweil Educational System's Kurzweil 1000 version 11, and ABISee's Eye-Pal and Zoom-Twix. Because my focus is high-volume print processing and editing, I will not discuss mobile devices such as the KNFB Reader Mobile since these are less suited to the task.

OpenBook and Kurzweil 1000 are fairly similar in intent and function; both work with mainstream flatbed scanners and offer powerful editing. Both have two OCR engines to choose from and can save files as audio. Both give users the option of downloading books from Bookshare.org, NLS Web Braille, etc.

The differences between Kurzweil 1000 and OpenBook are in some of the finer points. Kurzweil 1000 saves its audio as DAISY-format files, which can be a significant advantage in documents with good markup. It also opens files from a range of sources (though without the option of saving the files directly in the source technology); PAC Mate, Maestro, BookCourier, Book Port, Road Runner, Braille ’n Speak, and BrailleNote devices are all supported, though the Victor Reader Stream, the most popular book-reading device at the moment, is not.

Kurzweil 1000's optimized scanning function uses both OCR engines and configures other settings (brightness etc.) for best results. This is a helpful function. Settings can then be saved for later use on similar documents. Bookmarks, text notes, and annotations can be added to study material, enhancing this software's use to the student or researcher. Finally, the program also includes an option to recognize forms, though we had little success with this feature since the field-recognition in our tests was not reliable enough to be truly helpful.

OpenBook 8 allows users to save to and open files from PAC Mate, Braille Lite, Type Lite, Braille ’n Speak, Type ’n Speak, BrailleNote, and VoiceNote notetakers. In our tests, however, the BrailleNote files did not open in OpenBook. The highlighting option in OpenBook has three colors, and, when the user exports the highlights, he or she can opt to group them by color or leave them in text order. While the highlighting covers some of the same ground as the bookmarks and annotations in Kurzweil 1000, the highlighting makes creating summaries easy and is a little more intuitive than the parallel function in Kurzweil 1000.

ABISee's Eye-Pal and Zoom-Twix are fundamentally different from OpenBook and Kurzweil 1000, though they cover some of the same ground. Where the two programs discussed above are essentially software solutions using mainstream hardware, the ABISee products use a combination of custom hardware and software.

The Eye-Pal and the Zoom-Twix have the same basic setup, but their audiences differ. The Eye-Pal, with its Braille support and JAWS compatibility, is aimed at blind users, whereas the Zoom-Twix, with its extra camera for distance viewing, is intended for the low-vision user. The basic hardware is the same for both devices--a small camera on fold-out legs connected by USB cable to a computer, which makes it much easier to scan books because one can simply flip the pages rather than having to unbind the book or lift it off the scanner to turn the page. The camera is also a portable solution; unlike a scanner it folds up to fit in a custom bag. The Eye-Pal and Zoom-Twix are designed for fast scanning. The software bears this out; it has an automatic book-scanning function, which recognizes when you have turned a page and have stopped moving, then takes a picture. The camera setup and this automatic scanning yield the best results for fast scanning.

Once the scanning is complete with either of the ABISee devices, some more profound differences between them and OpenBook and Kurzweil 1000 are apparent. The ABISee software lets you save files as text, but any editing has to be done in a full-fledged editing program.

The bottom line for anyone choosing an OCR solution is this: if what you want is to be able to scan fast, the ABISee products will be your choice. If, on the other hand, you need to edit and reformat your files extensively, you are likely to prefer OpenBook or Kurzweil 1000.

The pricing and purchasing information for the products discussed in this article follows:

The purchase price for Kurzweil 1000 version 11 is $995. Contact Kurzweil Educational Systems at (800) 894-5374 or visit <www.kurzweiledu.com>.

OpenBook 8.0 is available for $995 from Freedom Scientific. You can contact them at (800) 444-4443 or visit <www.freedomscientific.com>.

The Eye-Pal at $2,000 and Zoom-Twix at $3,500 are sold by ABISee. Contact the company at (800) 681-5909. You can visit the Website at <www.abisee.com>.

If you have questions, contact me at <cvangerven@nfb.org> or at (410) 659-9314, ext. 2410.

(back) (contents) (next)