FineReader 6.0 Professional OCR Upgrade

Rating:
Summary: FineReader 6.0 - Best OCR software
Review: FineReader Pro 6.0, which I have run on Windows XP since March, is best of breed in batch processing, scanning, page analysis, OCR, editing, saving, stability, documentation, and customer service.

Batch: named set of pages processed together. Set the options once, and leave most of them, creating a new batch for a different set of options for resolution, patterns, etc.

Thumbnails are on the left, with page state indicators. Selected pages can be renumbered, saved, or deleted. Clicking a thumbnail opens the image, text, or both, each with its own zoom factor. A zoom pane shows a highly magnified view of the position.

FineReader supports PDF, BMP, JPEG, PCX, DCX, TIFF, and PNG image files. They are added to a batch by opening within FineReader, or by dropping one or more of them on a FineReader Window.

I set the scanning options to use the FineReader interface, which is much easier to use than the HP Precision Scan Pro interface.

FineReader scans single or multiple images at the click of a button. It deskews, orients, and despeckles images. If prompting for page number is selected, the user can specify the first page number, direction, and continuous or skip for a set of pages dropped on FineReader or scanned. A stack of two-sided pages can be placed in the ADF, scanned forward with skip, turned over, and scanned backward with skip to get all the pages in the proper order.

A batch is usually saved to different files, a few pages at a time, and I often save recognized pages in the batch that is currently doing background scanning and OCRing.

If the ADF is empty the user is prompted without stopping the background tasks. I can place a single page in the flat bed, a stack in the ADF, or cancel the scan. To scan multiple odd sized pages in the flat bed, uncheck the ADF option, and set the time between scans. Scanning the flat bed will continue until the Scan button is toggled.

Multiple instances of FineReader run concurrently. While one instance is scanning documents in the ADF and recognizing the scanned pages in a background thread, I edit and save previously recognized pages of another batch. There is little to do beyond feeding the ADF and starting Save operations, and the CPU is 100% busy whenever I have many pages in the ADF. I set FineReader's priority low, so other applications remain responsive.

FineReader shines for OCR. When I first started using FineReader 4.0, I compared it against Textbridge and Adobe Acrobat, and found that FineReader made very few recognition errors compared to the others. FineReader Pro 5.0 made very few recognition errors compared to 4.0 and compared to ReadIris Pro 7. For example, difficult letters like i, m, t were missed in FineReader 4.0 and rarely missed in 5.0. FineReader Pro 6.0 makes still fewer errors.

FineReader's recognition process has two parts: layout analysis, and character recognition. With the background recognition task, both parts are done automatically as pages are scanned or dropped. Even while background recognition is going on for new pages, manual layout can be done for other pages. Text, Table, and Picture blocks can be resized and reshaped. Blocks can be added or deleted. Block types can be changed. Deleting a block removes it from the saved output; changing to a picture block puts the image in the output, with no recognized text. Neither action requires re-recognizing the block, so, when I would rather have an image of a block than recognized text, I can quickly switch to a picture block.

Automatic layout analysis is usually accurate, but sometimes pages with hand-written information need to be adjusted so that all the handwriting is in picture blocks.

OCR errors take two forms: incompletely recognized characters, which may be recognized rightly or wrongly, and incorrectly completely recognized characters. FineReader 6.0 has almost no incorrectly completely recognized characters, and allows the few that slip past to be edited. FineReader 5.0 had quite a few more, and did not allow such errors to be edited and saved in PDFs, so I had ended up saving many more results in Image over Text format than I would have otherwise.

Pattern training is usually not necessary. It only becomes useful for highly unusual fonts, or for large sets of documents that can all be scanned with the same resolution and that contain the same few unusual fonts.

User patterns can be named and are saved with batches. When pattern training is selected, recognition stops at each uncertain character. The user can confirm, change, or skip. Character attributes, such as superscript, can be set.

FineReader 5.0 did not preserve the page layout during text editing, and did not allow completely recognized words to be edited. Both these problems have been corrected in 6.0.

A spell check box displays the word image in context, with wider context in the zoom window, and choices for resolution: accept; Add the word to the dictionary; Confirm the changes; skip.

Results can be saved as DOC, RTF, PDF, HTML, CSV, TXT, XLS, or DBF files, and various formatting options are supported. PDF formats are Text and pictures only, Page Image Only, Text over the page image, Text under the page image, with optional images instead of incompletely recognized text. Maximum resolution and JPEG quality can by specified.

FineReader is stable. It continues from where it left off when a jam is cleared. It does not get GPFs; it does not hang. If a TWAIN driver is cancelled, FR6 recovers gracefully, and reinitializes the TWAIN driver. There are never problems with task synchronization. The current state is always saved to disk, so that even a power failure does not cause loss of data.

FineReader's Help is excellent, well organized and complete, with an excellent search facility.

I deal with Abbyy Customer Support by email. I have had few problems, several questions, and many suggestions for product enhancements. I always get prompt, courteous, useful responses, and quick fixes for problems.

Rating:
Summary: dissapointing PDF
Review: I bought this product after talking to one of the Abbyy reps at a trade show in Las Vegas. My primary reason for buying was to turn documents into PDF files for repository storage. I also wanted to take some of my stored PDF documents and edit them in MS Word....conversion into PDF was ok (words were right, formatting was screwed up) but converting PDF into word was horrible....worked nothing like the demo they showed me

Rating:
Summary: dissapointing PDF
Review: I bought this product after talking to one of the Abbyy reps at a trade show in Las Vegas. My primary reason for buying was to turn documents into PDF files for repository storage. I also wanted to take some of my stored PDF documents and edit them in MS Word....conversion into PDF was ok (words were right, formatting was screwed up) but converting PDF into word was horrible....worked nothing like the demo they showed me

Rating:
Summary: Very much the best overall
Review: I do a great deal of OCR in my research and have scanned thousands of pages each with OmniPage Pro 12 (OP12) and FineReader 6 (FR6). I have also made extensive use of previous versions of both programs. I am happy to have both, because they excel at different things. For my work, however, if I had to choose just one it would have to be FR6. Others might have different preferences depending on what they do and what their equipment and software are.

The bulk of my scanning/OCR involves academic articles and historical materials. For the most part I produce PDF files, although I also scan some tables to produce spreadsheets and do some scanning to Word files. Depending on the quality of the original and my precise purpose I may make a PDF with an image and hidden text, an OCR text file, or an OCR text file with images of uncertain words. I use an HP 7450 scanner connected to a Windows 2000 system with a 1.8 GHz P4 processor and 512 MB of RAM.

For my purposes, OP12's outstanding feature is the quality of its grayscale and color scans. In fact, I even sometimes use it to produce images for processing by FR6. Generally speaking, the PDF image files produced by OP12 seem to run about 80% smaller than those that FR6 produces for equal text quality -- and better rendition of photos! This is not true for black-and-white (1 bit) files, where FR6 seems to have a slight edge. But when the material calls for image output I usually click on OP12.

OCR is another story, for several reasons. First of all, when the going gets tough, OP12 quits in a huff. It will suddenly crash no warning whatever. This seems to be OCR-related, but if it happens while scanning the chances of recovering your already-scanned work are poor. For this reason, I always scan and recognize separately with OP12, since then the crashes usually do not corrupt the scanned images. Depending on the complexity of the material, I may get a crash anywhere from one in every 20 to one in 100 pages.

Naturally, separate scanning and recognition slows the process down. On top of that, OP12 is very slow to start with, at least with "only" a 1.8GHz processor and 512 MB of RAM and all other applications closed. When I need fast results or cannot tolerate crashes I use FR6, which is distinctly faster and seems nearly bulletproof.

Moreover, when accuracy of scanning counts, OP12 is next to useless for my purposes. That's because it is very weak on anything but straight text. Superscripts all look like quotation marks to it and subscripts all come out as commas. It is also very poor with any sort of special symbols or equations. Nor is there any way to correct these mistakes in the editing process -- you're forced to edit the PDFs with Adobe Acrobat, a very slow and laborious process. If you have material with as many superscripts, subscripts, and special symbols as the typical academic article, it is really faster to retype it than to try to do it with OP12. FR6, by contrast, gives reasonably good accuracy with such material and makes it easy to correct the mistakes that do crop up.

In a surprising number of cases, OP12 will rotate the page so that the text is not upright and then proceed "recognize" it as garbage. FR6 is not immune to this, but does it significantly less often.

FR6 is sometimes wrong but never in doubt -- it has never reported being unable to complete OCR of a page, no matter how complex. OP12 is easily confused, especially when the page mixes text and tables, and then insists that you manually zone the page before it will proceed.

Both programs offer an "auto-special" completely automatic mode that will do a decent job on simple material (assuming that OP12 doesn't crash in the middle). When you need to customize settings, however, FR6 offers more range of choices. It also offers more flexibility in correcting recognition errors and in manual zoning, should that be necessary.

Surprisingly for a version 12, OP12 has a great many glitches, bugs, oddities, and time-wasting annoyances that make it seem more like an early beta. About 20% of the PDFs it produces are unreadable -- it's important always to check. The early FineReader versions were extremely rough, but FR6 is a very stable and finished product.

As I say, I don't regret the money I've spent on either of them. However, FR6 is more generally useful, faster, and trouble-free -- and significantly cheaper. If the maker of FR6, ABBYY, would fix their scanning (something I've suggested to them several times) it would clearly be the preferable program. As things are, OP12 fills some needs better.

Will O'Neil

Rating:
Summary: Great Softwar - Buy The Upgrade
Review: This package works extremely smoothly and accurately. The way it is priced is confusing however (more below). FineReader is not only easy to use and accurate, coping with curved lines of text coming from scanned book pages, recognizing tables, fonts and preserving layout of the scanned image, but it also organizes your work, saves images in batches, and keeps track of which image you have proofread, modified and saved. While you're proofreading you can see the part of the image about which there are questions. After the briefest of practice sessions I dove into a 700-page project and it went totally smoothly. The Abbyy website also offers you a fully functional free trial version that you can download.

Regarding price, if you have any -- that is any -- OCR software already installed on your computer -- such as came with your scanner, the Abbyy website informs you that you are entitled to the "upgrade" version of this package, which costs only half as much as the "full" version! I had thought that "upgrade" was only useful for those with previous versions of FineReader, but this is not the case. I wonder how many people have paid the extra $150 not knowing this fact.

Rating:
Summary: Great Softwar - Buy The Upgrade
Review: This package works extremely smoothly and accurately. The way it is priced is confusing however (more below). FineReader is not only easy to use and accurate, coping with curved lines of text coming from scanned book pages, recognizing tables, fonts and preserving layout of the scanned image, but it also organizes your work, saves images in batches, and keeps track of which image you have proofread, modified and saved. While you're proofreading you can see the part of the image about which there are questions. After the briefest of practice sessions I dove into a 700-page project and it went totally smoothly. The Abbyy website also offers you a fully functional free trial version that you can download.

Regarding price, if you have any -- that is any -- OCR software already installed on your computer -- such as came with your scanner, the Abbyy website informs you that you are entitled to the "upgrade" version of this package, which costs only half as much as the "full" version! I had thought that "upgrade" was only useful for those with previous versions of FineReader, but this is not the case. I wonder how many people have paid the extra $...not knowing this fact.