Applelinks.com: OCR Review

ScanSoft owns both OmniPage Pro and TextBridge. This kind of a situation is potentially bad if ScanSoft chooses to remove one or the other of these programs, but as it is, they are as different as night and day and both good on their own. But as things stand, both exist and a duel seemed in order.

Part I, What does OCR mean?

When you scan a page of text, you have a picture of the scanned page and nothing more. You cannot edit that page nor can you copy from that page. However, if you own some OCR software, you can convert that picture of text into editable text that you have full use of.

OCR stands for Optical Character Recognition. That means it can (with the aid of a scanner) take the text from a piece of paper and "transform" it to editable text on your computer in a word processing program such as Word. The key concern here is "can the program do it faster than you could do it by retyping in the same material?" This last issue is fundamental because consider the process of using OCR:

If all you have to convert is a paragraph or two, you could be done by the time you got to step 4. If you have a page, it could be a wash (depending on your typing speed and quality), but over a page of text, OCR software will invariably save you time. However, that depends on the quality of the conversion and the nature of the text being converted. More on that later.

Part II, The program interface.

I cannot begin to tell you how much I dislike the TextBridge user interface. The entire process from text capture to looking at your document is as non-intuitive as any program I've seen. Also, there are features of the program that are literally hidden from view and unless you have been told (or read) how to make some of the features visible, you could go your entire life never even knowing they existed. In Figure 1, you see the basic TextBridge interface.

Figure 1

If you move your mouse over the double line on the bottom, you can drag it down to display the other features/preferences. (See Figure 2) Please note that in Figure 1 there is no visual indication that there are extra features hidden from view--you have to know to lower the double line to even consider lowering the double line.

Figure 2

With TextBridge, the user clicks the "Go" button and before anything else is done, saves the material in the format desired--then scans each page. If you click the preview button (the one to the left of the "man-at-the-blackboard" button) before you scan a page, you will get one more version of the control strip seen in Figure 3:

Figure 3

Here, the "T" means you can "zone out" blocks of text you want to be OCRd, the "Mona Lisa" lets you block out pictures you don't want the program to try and convert into text but stay as pictures, and the "negate" button lets you block out areas you don't want any action to be taken. If you don't have the Preview button pressed, you are dependent upon TextBridge to figure out what you have and how it is to be laid out.

Each time you click the Green Arrow, you are offered the option of continuing or ending your session.

Figure 4

With OmniPage, the user (1) "grabs" the page(s) by clicking on the scanner, defines the regions (using the Tools seen in Figure 5--pretty much the same rules as mentioned for TextBridge), (2) proofs the page(s), and (3) saves the final output.With OmniPage, one saves after having completed the scanning.

[On a side note, when I think back to OCR conversion on my old IIci, it is

Figure 5

wonderful to watch the speed at which both TextBridge and OmniPage reads through the text for conversion--it is a wonderful treat (this is on a beige G3, 233).]

Part III, The quality of the output

Regardless of the interface, what the bottom line must be is "how good is the final result?" Here you are in good hands either way--with very important differences.

In a test on a 3000+ word document in a fairly small font (about size 7 points) with super- and sub- scripting and occasional scientific formulas, both had a comparable number of mistakes: OmniPage (26) and TextBridge (23). However, While OmniPage could not see a super-/sub-script to save its life, it had no actual "mistakes" in reading the text. TextBridge had three mistakes in text reading and two mistakes calling normal styled text into bold. Although TextBridge could read the super-/sub-scripted text, it didn't do a very good job. (However, due to the size of the text it was working with (about 7 point with an approximate size 4 point super-/sub-script text, I really cannot be too upset--TextBridge does understand super-/sub-script text which is more than OmniPage can do.)

In addition, TextBridge can re-attach hyphenated words. OmniPage split hyphenated words leaving the "hyphen" as a lowly tail. (In MS Word, they are carried over with a "soft" hyphen. (That is they still have the hyphen if you are looking with the "show Paragraph marks" on and are not seen with the "show Paragraph marks" off.)

Another feature of TextBridge is that it can replicate pages. That is, if you have a page with text and pictures, TextBridge can reproduce that page with all the text and pictures (mostly) in exactly the same position as they held on the original page. Also, TextBridge is fully Apple Scriptable.

On the other hand, OmniPage can self-straighten pages which is fantastic when you are trying to scan pages out of a book and can't get the book to lie square on the scanner screen. Also, OmniPage can rotate pages 90° either way or 180°--thus, a page's original orientation on a scanner is not critical. In addition, OmniPage shows a thumbnail of all the pages you have scanned and you can re-arrange the order, and/or delete pages as needed before you start the OCR process.

If you have text that is somewhat stylized, with TextBridge one can "teach" the program how to better read the nuances of the unique font. That's what the "man-at-the-blackboard" button is for in TextBridge Figures 1, 2, or 3. These can be saved for future use when using the same font at a future time.

Part IV, The final output

A bit over four years ago I reviewed TextBridge versus TypeReader (a wonderful OCR program that is no longer available for the Mac). TextBridge was so bad in everything it did, I couldn't see how it had any future. I'm glad that Xerox/ScanSoft kept at the program as it is now a very good program despite the fact that it has an interface that sucks. Despite this interface, TextBridge can reproduce page appearance, see super-/sub-scripts, understands hyphens, and perform an excellent job of text recognition.

For years, OmniPage has been the leader in OCR software. Now it is being shoved around by an equally powerful TextBridge. Sometime in the next several months OmniPage will be coming out with a new version and I await it very anxiously to see how the scan-wars turn out.

Not wanting to bore you too much with the corporate world, a (very) short history is kinda interesting here. ScanSoft (formally a subsidiary of Xerox) merged with Visioneer in March 2, 1999. Coming from Xerox, ScanSoft already owned TextBridge. By the end of April 2000, ScanSoft completed the purchase of Caere Corp., creators of OmniPage Pro. Thus the two major OCR software programs for the Mac are now owned by the same company.

Looking at what has happened with the various repair programs purchased by Symantic, the future doesn't look good for one of these programs. Here is one case where I really hope that ScanSoft doesn't follow tradition.

OmniPage versus TextBridge Or ScanSoft Versus ScanSoft

Part I, What does OCR mean?

Part II, The program interface.

Part III, The quality of the output

Part IV, The final output

OmniPage

OmniPage versus TextBridge
Or
ScanSoft Versus ScanSoft