IrisPen II Executive Asian (handheld scanner and software)
Working in Japan in 2002, one still often comes across amazingly low-tech solutions. One of my favorites (in retrospect) was a manager at an office where I worked who laboriously typed his drafts of English letters (IN ALL CAPITAL LETTERS) on a small laptop computer, printed them out on his own personal inkjet, and then asked me to use my computer to improve his English, based on the printout. When I suggested that a floppy disk rather than a sliced dead tree might be of assistance (needless to say, the office had no networking at this time), he seemed genuinely astonished that this might be more efficient.
Even though Japan has become much more wired of late, there are often times when we writers, translators, and editors need to extract information from hard copy. Last year’s company brochure which needs to be updated may well exist only in hard copy, or else in some weird and wonderful Japanese data format that seemingly cannot be converted without specialist expensive add-ins. The facts and figures presented at a meeting cannot be given in electronic format “for security reasons”. The old document on which you have to base your new document was produced by an old-style phototypesetter by a printing company that has subsequently gone bankrupt. And so on and so forth. You are encouraged to find your own examples.
Why type, when a computer can do it for you?
Typing in from hard copy is typically not enjoyable or accurate. Although OCR solutions (for European languages and Japanese) have existed for some time, typically they require a whole page to be scanned at least once, even if the only required information is a name and address, etc. or a few figures for a report.
What we really need in many cases is the electronic equivalent of scissors and paste, that can copy and paste small chunks from the printed text into your computer.
When I discovered a handheld scanner and associated specialist OCR software on the web, I was quite interested, especially as it promised to do Japanese. The company is I.R.I.S. at www.irislink.com, a Belgian company that probably takes advantage of the paper mountain at the EU headquarters and develops tools to help to reduce this bureaucratic nightmare.
To pique my interest still further, the IRISPen II Asian Executive also claimed that it would work on Windows as well as Mac OS X (and Mac OS 9, but I don’t use Mac OS 9). Note that the Mac OS versions do not read Japanese, Chinese (either flavor) or Korean, but do everything else described here.
I have a few needs for inputting European languages other than English—such as Finnish safety warnings, which sometimes need to find their way into English manuals—but the fact that the software claimed to support more than fifty European languages was a plus point (my needs for Byelorussian and Kurdish support are limited, though).
As is the case with most products, I like to know a little more than the Web site provides before I commit my money, and I found the pre-sales support to be at least adequate, answering my questions in a timely and accurate fashion.
So I used my credit card to the tune of EUR300 or so (plus DHL shipping) for the full, deluxe, all bells and whistles version of the scanner plus software IRISPen package, and a week later, the box was in my hot little hands (Japanese customs had opened it and discovered hardware inside, and therefore charged me ¥2000 duty).
Installation
Within 10 minutes of opening the package I was successfully scanning and reading text into my iBook computer (OS X). The CD-ROM provides an easy VISE Installer. No reboot required, and the USB scanner just goes straight in. Because under OS X the software emulates a keyboard (or Apple Events, or scans text to clipboard), you can use it with any software package.
Fonts are not matched when the text is entered into the application (unlike, say TextBridge or OmniPage) and the text comes into the application in the currently-set font, but the speed of recognition is so fast I couldn’t believe it was happening.
The pen scanner itself takes a little bit of getting used to, especially if you’re a left-hander (but there is a left-hander’s option). You hold the scanner a little like a fat highlighter, and drag it to from left to right along one line of text. To ensure that you hold it correctly, the roller at the tip acts as a switch, and the tip illuminates, with a small window displaying the text that you are currently scanning. If you find yourself scanning small text with many lines, the software appears to be smart enough to detect which line is in the center, and it rejects the surrounding text. Like many techniques, this demands a little bit of practice before it becomes perfect, but it has taken me under 20 minutes to get to the stage where I scan the line that I want well over 90% of the time. It is also possible to use an on-screen window showing the area which has just been scanned in order to correct your aim.
Provided together with the documentation and software is a multilingual test sheet with a variety of fonts ranging from 8 point (this is about the minimum type size that is realistically recognizable by the software—too small for many business cards, more’s the pity) ranging up to 22 point, which is the largest type size (determined by the width of the scanning head). To make life even easier, it is also possible to restrict the recognized characters to alphabetic or numeric characters (or even to bar codes—the type of bar code that is recognized is automatically detected by the software).
Because the scanner is gray-scale, it is possible to scan color documents where the text is not black and the background is not white, but there are two colors which cannot be recognized by the scanner: red and yellow, which means that text in or on these colors cannot be recognized. Still, it would be unreasonable to expect perfection from a unit of this price.
How good is it?
The accuracy is claimed to be over 99%. In practice, I find it to be a little higher, but then I think that the manufacturers are being somewhat conservative in their claims. In order to help with the sensible recognition of text, you specify the language in which you scan, and the software compares its output to a lexicon, ensuring that as far as possible you do not get gibberish and meaningless combinations of characters appearing in your document.
Also, since with a pen scanner you must keep your eyes firmly on the document that you are scanning (unlike a flatbed scanner), you can set the software to read back what you have just scanned, using the speech synthesis software on Mac OS X or the software speech synthesis provided by the scanning software (Windows). This means that if you are scanning a number of lines, and your scan position is off by one line it is obvious immediately that you have done it, if you are scanning the line at a time.
The reason for that last caveat is that it is possible to set up a multiline scan mode, where each line is buffered and added to the last scanned text until the button on the pen scanner is clicked, flushing the buffer, and sending the scanned data to the target application. Spaces are automatically added and hyphens are automatically removed in this mode, making it possible to scan very short lines of text in multicolumn layouts, for example.
…….’” Accuracy is claimed to be over 99%. In practice, I find it to be a little higher, but then I think that the manufacturers are being somewhat conservative in their claims. In order to help with the sensible recognition of text, you specify the language in which you scan, and the software compares its output to a lexicon, ensuring that as far as possible you do not get gibberish and meaningless combinations of characters appearing in your document.
The example above was input using the multiline mode (see below) scanning a paragraph of this review after it had been printed out in 12 point Times on a laser printer. The garbage at the start was due to my misalignment of the first word of the first line. I have not touched the rest of the text—this is just as it came out of the scanning software.
Another function which is claimed, but I have as yet been unable to implement properly under Mac OS X (it does seem to be working under Windows, however) is the ability to scan rows of figures separated by vertical bars and automatically insert a character such as a tab character between the groups of figures. Such an application is extremely useful when entering figures into a spreadsheet.
The Japanese mode, as explained earlier, is only available on a Windows version of the software. Although I have not tested this is fully as I have tested the European language scanning, it seems to work very well with a high degree of accuracy, and seems almost as fast as the non-Japanese European languages scans (my wife stood open-mouthed when she saw this at first and I have had to install the software on her computer—the license allows this, by the way). Even vertical Japanese (tategaki) is supported in the latest version, meaning that it is even possible to scan Japanese newspapers and magazines printed in this way.
And what’s wrong with it?
One shortcoming of this system is that the pen scanner takes a little time to get used to compared with other scanning methods such as sheet-feeder or flatbed scanners but the technique, once acquired seems to be solid and work well.
There is also no learning mode in which it is possible to train the software to recognize the particular styles of typography or handwriting (there is a mode in which the software can recognize handwritten figures, but not letters). On the other hand, it is possible to set up a substitution table in which consistently misrecognized characters can be substituted with the correct character. In fact, there are number of ways in which the parameters of the software can be tailored to individual needs, and the settings can be set and stored as default options for later recall when a similar scan session is carried out.
Right now, the software has yet to pay for itself in terms of my time saved, but I think it is only a matter of a few weeks at most before I can fully justify the cost of this unit.
Recommended for:
- anyone (writers, analysts, translators, editors) who has to input snippets of text from hard copy.
Not recommended for:
- large bulk OCR work (I.R.I.S. does make OCR software for this kind of work)
- Left-handers may find it a little tricky, despite the laudable attempt to cater for them
- Anyone who needs to scan Japanese and uses exclusively Macintosh systems
I.R.I.S. Europe
Rue du Bosquet 10
348 Louvain-la-Neuve
Belgium
Phone: +32-10-4513 64
Fax: +32-10-45 34 43
I.R.I.S. Inc.
1600 N.W. Boca Raton Bld, suite 20
Boca Raton, FL 33432
Phone: +1-561-395 7831
Fax: +1-561-347 6267
support@irislink.com
http://www.irislink.com
Comments
Next entry: Google Pocket Guide
Previous entry: Check with Beckwith