Product Description

ABBYY FineReader Engine 11 for Mac is a comprehensive software development kit (SDK) for integrating ABBYY's multilingual OCR, OBR, BCR, document classification, document imaging, document conversion and PDF conversion technologies into applications for Mac operating systems.

What Is New

New Usage Scenarios

Feature	Description	Benefits
Automatic document classification	The task of document classification is to assign a document to some category on base of its content. ABBYY FineReader Engine 11 provides an API for automatic document classification that enables applications to categorize and sort batches of documents by predefined document classes.   Classification in FineReader Engine 11 can be performed in two modes (classification profiles): Maximum Speed. This mode is useful for documents that contain not much text, and the difference between classes is visible in the appearance of the documents. It uses: Image pattern (black pixels location template) OCRed text analysis: Title text Maximum accuracy. This mode is useful for documents that contain a lot of text, and the difference between classes can be determined only when text content is taken into account. It uses: OCRed text analysis: Full-text The “Maximum speed” mode provides classification speed from 3 to 10 times faster than “Maximum accuracy” mode. The classification results are: Detected category of the document Probability that a document belongs to a category Sample usage scenarios: Archiving: Sorting documents by type for electronic archive creation Mailrooms and Workflow Automation: According to document class detected some further actions can be initiated Batch Processing: Document separation BPO: Pre-sort documents for further processing Banking/Insurance: Verification of document set completeness OEM: Smart MFP/scanner interfaces suggesting typical actions	Enables workflow automation Boost productivity and reduce costs by eliminating manual pre-sorting It’s easy, no specific knowledge needed Easily adjustable Universal Easy to integrate
Business Card Recognition	Business card recognition technology now is integrated in FineReader Engine 11. The API provides a full set of features for business cards processing.   The following fields can be extracted: Personal name Company name Position in the company Company address Phone number Fax Mobile phone number E-mail Web site Export to vCard format: Recognized data can be saved in vCard format Business Card reading in 27 languages: ChinesePRC, ChineseTaiwan, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Indonesian, Italian, Japanese, Korean, Norwegian, NorwegianBokmal, NorwegianNynorsk, Polish, PortugueseBrazilian, PortugueseStandard, Russian, Spanish, Swedish, Turkish, Ukrainian Auto-splitting of multiple business cards: Multiple business cards can be detected on a page Multiple business cards scanned on one page can be split	Fast and easy way to transfer contacts Superior accuracy Add more value to your application A must-have add-on for any CRM system

OCR Improvements

Feature	Description	Benefits
Arabic OCR	FineReader Engine 11 includes Arabic OCR technology.	Expand your business in Arabic markets Put your finger on the pulse of Arabic world
CJK OCR	CJK OCR is now available for Mac users.	Enter to local Oriental markets to compete with domestic OCR vendors
New languages for OCR	FineReader Engine 11 for Mac supports 27 new recognition languages.	Widest language support available in the industry – 202 languages for OCR
Maxicode barcode support	Maxicode barcode is used for tracking shipment of packages.
USPS 4CB barcode type	USPS 4CB or IMB is a barcode used by USA post office.
Receipt recognition	This type of text is designed for recognizing sales receipts, invoices, etc. Unlike other types, it is not concerned with the actual font of the text. Rather, it tells the recognizer that there may be text of low quality, mostly in monospaced or normal font. The typical receipt text can look like this:
New mode for low resolution scans	Special recognition mode for low quality documents provides 20% higher accuracy.	Now you can recognize the low quality documents that caused too many OCR errors before

New Image Preprocessing Tools

Feature	Description	Benefits
Camera OCR   Preprocessing features for photographed documents	New version of FineReader Engine includes the following new and improved preprocessing features: Geometrical distortions correction                                    (not only trapezium distortions as in previous version, but any type of geometrical distortions), Auto-cropping Background lightening ISO noise removal	Better recognition results Allows to produce high quality searchable PDF with excellent appearance
Auto-splitting of double-page spread	Books are usually scanned as double-page spreads. This produces some difficulties for recognition (curved lines, scanning shadows). In the output document it is usually better to have the book page-by-page. New version of FineReader Engine 11 can perform page splitting automatically. This means higher effectiveness of image preprocessing (curved lines correction, scanning shadows removal).	Better appearance of output document (page-by-page).
New image preprocessing methods	RemoveNoise Method. This method reduces the noise on the image. EnhanceLocalContrast Method. This method increases the local contrast of the image. Such preprocessing may increase recognition quality of low contrast images.	Better recognition results
New binarization	Previous OCR SDK version provided very high quality of binarization, but in some the most difficult cases it could commit errors and losses of information. New binarization technology ensures the whole text retention and prevents information losses even in difficult cases.	New binarization provides outstanding quality and guarantees keeping of valuable information resulted in better OCR accuracy
Color stamps and marks filtering	If there are some stamps or marks, made by pen, marker on document image they usually interfere text and decrease OCR results. That is why ABBYY FineReader Engine 10 includes special feature for color marks and stamps filtering and improving recognition accuracy.	An excellent feature for data capture systems allows preventing data losses from fields covered by stamps and color marks

Performance Improvements / Shortened Development Curve

Export Improvements

Feature	Description	Benefits
Improved PDF Export
Faster PDF Export	Export to PDF now is up to 2-3 times faster than in previous version of FineReader Engine 8 for Mac.	Converting images into searchable PDF is one of the most needed scenarios on the market.
PDF export profiles	ABBYY FineReader Engine 11 for Mac provides predefined profiles with optimal values for popular export variants: MaxQuality Balanced MinSize MaxSpeed With predefined PDF export profiles you automatically set optimal values for particular task.
Higher quality of PDF MRC	PDF MRC improvements include: higher background image compression, contrast elements stay in foreground. Higher background image compression reduces the size of output PDF MRC file for up to 50%.
Stamps and written notes processing for PDF MRC * * still under development	Stamps and written notes can be placed to any part of documents and lay over sensitive information, which is needed to be extracted. Such marks are merged with the text during binarization and sensitive information can be lost. FineReader Engine 11 provides special preprocessing mode for such cases. The idea of this preprocessing mode is that an image is split into two layers: color and black-and-white. Black-and-white layer is used for recognition, while the color layer is passed to export without modifications and it will stay in foreground of resulting PDF MRC file instead of appearing in background as blurred elements. As a result, output PDF has high-level quality and compression. Text and separators are compressed with suitable black-and-white codec, while color layer - with color codec.
An ability to create linear or non-linear PDF files	Linear PDF files have internal data arranged in a page order. A page of a linear PDF file can be read in a web browser plug-in without waiting for the whole file to be downloaded. Non-linear PDF have the data necessary to assemble a document page scattered through the whole file. New PDF export parameters include new option which specifies whether a linear PDF file should be created.
Other Export Improvements
Docx and Xlsx Export Improvements* * based on ABBYY internal test results	Docx improvements: Better pages number detection Better font detection: Font type , Bold, Italic, Header Better text wrapping Better page orientation detection Xlsx improvements: Better cells detection Better columns’ width detection
Excel export improvements* * still under development	FineReader Engine 11 allows retaining formatting of all data in the tables exported to Excel, including numbers: Bold font style Font colors
Export to XPS* * still under development	XPS (XML Paper Specification) format is based on XML. As PDF format, it provides device-independent document appearance. XPS also comes handy when one doesn't have a printer installed and the XPS virtual printer allows to save the document in "ready-to-print" original format for later printing.
Export to memory* * still under development	FineReader Engine 11 will be able to save recognized documents not only on disk, but into a file stream.	Increased security for your confidential data
Extended ABBYY XML	Now there is an ability to save paragraph style and roles into output XML. This can be useful to identify the role of a paragraph, e.g. to detect running titles and footnotes.
Other	Recreation of the logical structure of a document is an option during export to RTF, DOCX, and HTML formats. An ability to save information of paragraph styles and roles in XML file. New color settings for embedded pictures in RTF, DOCX, PPTX, HTML, EPUB, and FB2 formats.

Useful and Clear Developer’s Documentation

New Licensing

Other Improvements

New Licensing

Other Improvements

Upgrade From Previous Version

Installing on the same machine

ABBYY FineReader Engine 11 works with any previous ABBYY FineReader Engine major version installed on the same PC if products’ installation folders are different.

Compatibility of protection keys

ABBYY FineReader Engine 11 requires valid serial number for functioning.

Using source code for previous version

ABBYY FineReader Engine 11 has certain incompatibilities with API implemented in the previous versions described in “ABBYY FineReader Engine 11 and 8.0 Compatibility” section of the product Help file. Every existing customer willing to upgrade his copy of ABBYY FineReader Engine should read the article first.

Compatibility Issues With Version 8

Licensed 3rd-party Software

This version uses several licensed 3^rd-party libraries. They enable the product with useful functionality and require us to add certain acknowledgements and items in the product documentation and/or LA.

JPEG 2000 Kakadu library

11^th version uses JPEG 2000 Kakadu library for saving image files in JPEG 2000 format, or to export to PDF format with embedded JPEG 2000 pictures. That obliges us to specify certain copyrights in the product documentation:

• Working with JPEG2000 image format: Portions of this software are copyright ©2011 University of New South Wales All rights reserved.

Components Delivery

FTP delivery

FTP delivery is mostly used for trial versions but can be also used for usual sales. It includes:

1. FTP address of the Distribution Pack.
2. Serial Number.

DVD box

It includes:

1. Common DVD Box with DVD Box Cover
2. CD/DVD with the Distribution Pack copy and the CD/DVD Label.
3. Serial Number.

Distribution Components

Documentation

Sample images

There are prepared sample images for demonstration of basic scenarios and advanced technologies.

The distribution DVD contains all images in the following folder:

• DVD .\Samples

Code samples

The distribution contains samples described below in the following folder:

• DVD .\Samples

Samples for Developers

Samples for technology advantages demonstration

Licensing Model and Parameters

Licensing model and parameters are the same as for FineReader Engine 11 for Windows with the following exceptions:

• Hardware licenses are not supported
• The following Functional License Limitations are not available:
  • Data Capture (ICR/OMR): ICR, Cyrillic ICR, OMR
  • Visual Components

See details in the similar document for Windows version.

Protection Key Types, Activation, Deactivation, Registration

Supported Protection Types

1. Software protection keys:
   • Open (no activation).

License Limitations

Page Counter

Productivity Licensing means counting of processed pages or characters.

Engine treats a page as processed upon a call of the following methods:

• IFRDocument::
  • Analyze
  • AnalyzeAndRecognize
  • AnalyzeAndRecognizePages
  • AnalyzePages
  • Process
  • Recognize
  • RecognizePages
• IFRPage::
  • Analyze
  • AnalyzeAndRecognize
  • AnalyzeRegion
  • AnalyzeTable
  • Recognize
  • RecognizeBlocks
  • ExtractBarcodes
• IDocumentAnalyzer::
  • AnalyzePage
  • AnalyzePages
  • AnalyzeRegion
  • AnalyzeTable
  • RecognizePage
  • RecognizePages
  • RecognizeBlocks
  • AnalyzeAndRecognizePage
  • AnalyzeAndRecognizePages
  • RecognizeImageDocumentAsPlainText
  • ExtractBarcodes
• IEngine::
  • RecognizeImageFile
  • RecognizeImageAsPlainText
  • RecognizeImageDocumentAsPlainText
  • AnalyzePage
  • AnalyzePages
  • RecognizePage
  • RecognizePages
  • AnalyzeAndRecognizePage
  • AnalyzeAndRecognizePages

A page counter is increased by one only once for the same ImageDocument object regardless to how many times one uses analysis, recognition, or exporting methods with that object.

In case of character counter, the following methods have no effect:

• IDocumentAnalyzer::
  • AnalyzePage
  • AnalyzePages
  • AnalyzeRegion
  • AnalyzeTable
• IEngine::
  • AnalyzePage
  • AnalyzePages
• IFRDocument::
  • Analyze
  • AnalyzePages
• IFRPage::
  • Analyze
  • AnalyzeRegion
  • AnalyzeTable

Supported Languages, Types and Formats

Supported Recognition Languages

OCR Languages

Classification languages

FineReader Engine supports all languages for classification.

BCR languages

27 languages including:

• 4 hieroglyphic languages (CJK) - Chinese Traditional (Taiwan), Chinese Simplified (PRC), Japanese, Korean

Supported barcode types

Supported text types and field marking types

OCR text types

1. Normal
2. Fax
3. Typewriter
4. Matrix
5. OCR_A
6. OCR_B
7. MICR_E13B
8. MICR_CMC7
9. Fraktur/Gothic (available only under FineReader XIX add-on)
10. Receipt

Field marking types

1. Simple Text
2. Underlined Text
3. Text in Frame
4. Grey Boxes
5. Char Box Series
6. Simple Comb
7. Comb in Frame
8. Partitioned Frame

Supported Import and Export Formats

Supported import formats

Supported export formats

1. RTF
2. Microsoft Office file formats:
   • DOCX
   • XLS/XLSX
   • PPTX
3. PDF file formats
   • PDF
   • PDF/A (1b, 1a, and 2a and 2u)
   • MRC (Mixed Raster Content) for both PDF and PDF/A
4. HTML
5. TXT/CSV
6. ABBYY XML
7. EPUB, ALTO, FB2
8. ODT
9. vCard — for export of business cards only

GetEngineObject Function Changes

The functions and methods, which load the Engine object, have their syntax changed:

The GetEngineObject function and IEngineLoader::GetEngineObject method do no longer have the parameters for Open License. To use Open License, one should use the GetEngineObjectEx function or IEngineLoader::GetEngineObjectEx method.

The GetEngineObjectEx function and IEngineLoader::GetEngineObjectEx method have one more additional parameter, which specifies whether CPU cores should be used in shared mode.

Full Native 64-bit Support

FineReader Engine 11 provides 64-bit versions of libraries, including libraries for classification and BCR. Details on the distribution package can be found in the Help file.

Classification

Classification is a new feature in FineReader Engine 11. You can find the detailed description of the feature in marketing materials of the product. Below are several implementation details.

Classification modes

Classification in FineReader Engine 11 can be performed in two modes:

• Fast. This mode is useful for documents that contain not much text, and the difference between classes is visible in the appearance of the documents. This mode uses image pattern (black pixels location template) and recognized titles of a document for classification.
• Quality. This mode is useful for documents that contain a lot of text, and the difference between classes can be determined only when text content is taken into account. This mode uses full text OCR for classification.

The mode is specified in IClassificationParams::ClassificationMode property.

Classification confidence

The results of classification provide information both on the detected category of the document and the confidence, probability that a document belongs to this category (IClassificationClass::Confidence).  There is also a flag (IClassificationClasses::IsSuspicious) indicating whether classification of a document was uncertain, e.g. if classification detected two classes with equal confidence for one document. One may use these values to determine the way of further processing for the classified documents, for example, re-classify some of the documents manually, if necessary.

FlexiFormsDA and FullTextIndexDA options in API and licensing

Starting from FREngine 11, DA for Invoices and DA for Full-text Indexing add-ons are free of charge and removed from licensing scheme. This is because of the following reasons:

•  “TextExtraction” and “DocumentArchiving” predefined profiles require these add-ons. That is because the add-on modules were initially designed for those usage scenarios and therefore were included into corresponding profiles. Without add-ons Engine performs not optimally in those scenarios.
• Indexing and text extraction years ago were ‘additional’ usage scenarios for FineReader Engine while ‘document conversion’ was main one. It seems that all three scenarios are basic now for Engine.

Changes in licensing required additional changes in API:

• We declare the FlexiFormsDA and FillTextIndexDA properties as obsolete. These properties will be removed in the next version of the product.
• The old FlexiFormsDA property is replaced with IPageAnalysisParams::EnableTextExtractionMode and IObjectsExtractionParams::EnableAggressiveTextExtraction properties.
• The old FullTextIndexDA property is replaced with IObjectsExtractionParams::DetectTextOnPictures property.

CJK doesn’t Use User Patterns

CJK recognizer does not use cache so it is impossible to teach it with user patterns. The same situation is in ABBYY FineReader Engine 8.

Software and Hardware Requirements

• Operating System: Mac OS X (10.6.x, 10.7.x, 10.8.x)
• Memory requirements
• Hard disk space requirements

New Features and Improvements

Support for Russian with Accents language

Russian with Accents is an ordinary Russian language where tone stress is written under a character, sort of an alternative way of writing a character. This alternative writing is supported in FRE 11 as a separate language: RussianWithAccent.

Support for Russian Old Spelling language

FRE 11 supports OCR in Russian language used before the Russian Revolution in 1917. This language includes few outdated characters and slightly different morphology.

Language alphabet:

Support for FullAscii and Code32 1D barcodes

The following new barcode types are supported in this release:

Both types are supported in automatic barcode location and type detection.

Re-formatted Distribution.csv file for easier runtime file list composing

Well known Distribution.csv file got new structure bringing ease into a process of runtime distribution file list composing. It is also suitable for making automated (script) procedures.

The file contains the following data (columns):

1. Stage — the stage of working with FineReader Engine which your application uses.
2. Part — the way in which you are going to use this stage. For example, the Opening stage includes Scanning and Pdf parts. If this field is empty, the file is needed for the working stage in general. Do not filter the blank values out.
3. Details — further specific information about the operations in which the file is used.
4. x64/x86 — the operating system architecture. Again, the files marked "x64,x86" are necessary for both.
5. RequiredByModule — the values in this column are equal to Stage.Part.Details, and there is no need to filter this column if the first three have been specified correctly. But it can be used to check which modules have been included.
6. RequiredByInterfaceLanguage — the interface language for which the file is necessary. The files marked "Any" are necessary independent of language settings.
7. RequiredByRecognitionLanguage — the recognition language for working with which the file is necessary. The files marked "Any" are necessary independent of recognition language.
8. Optional — specifies if the file is necessary for the module functionality. If the value is No, this file must be included in your distribution kit. The value can be set to Yes in the following cases:
   1. the file is language-specific. Include it if you need this language (consult columns 6 and 7).
   2. the functionality for which this file is responsible is not always necessary. For example, it can be used for opening images in a specific format. Consult ABBYY FineReader Engine Distribution Kit for further information about this file which will help you decide if you need it.

Finally, you receive the list of files. In the last three columns you will find the information about these files' location and size:

1. Path — file path in the root installation folder.
2. FileName — file name.
3. Size — file size in bytes.

New export formats are supported in CLI sample: ALTO, EBook

CLI sample was updated to support additional export formats: ALTO and EBook.

JBIG2 lossless compression is supported

This release includes new compression format for black and white images – JBIG 2 lossless.

It is useful if somebody works with low quality b/w images where lossy format may lead to similar characters substitution, e.g. “6” to “8” and vice versa.

This format is available for choosing in PDFPictureCompressionParams::BwPictureFormats.

Improvements in MRC visual quality

It is possible to use black and white image prepared during binarization process as a source for MRC mask. In this case all contrast graphical elements will be saved into PDF MRC file with the highest quality.

This technique allows saving handwritten data, stamp details, signatures, etc. in a foreground layer with high quality (low compression) and keep a document look as close to original as possible. Previously enumerated objects were placed into background layer with high compression.

The following PDF export parameter enables the feature: PDFMRCParams::UseBwImageAsTextMask.

Method for checking if a page is empty

FRPage::IsEmpty() method checks if the page is empty. It uses the same parameters as analysis methods to find out if the page contains any relevant objects, for example text, tables, or pictures.

If there is a high probability that some of your pages contain only barcodes, and they are relevant for your procedure, set the boolean NeedCheckBarcodes parameter to TRUE to help the method detect barcode-only pages.

This method is useful in a batch scanning scenario when there is a need to separate scanned images flow into documents. In this scenario a batch contains separating paper sheets which are blank or with certain barcodes printed on them.

Aggressive table detection mode

PageAnalysisParams::AggressiveTableDetection property manages the table detection mode. If you set it to TRUE, FineReader Engine tries to find as many tables as possible on the page. This setting is recommended only for the documents which contain a lot of tables.

This property is FALSE by default.

Predefined processing profile for writing highly compressed image-only PDFs

The Engine got new processing profile “HighCompressedImageOnlyPdf” for creating highly compressed image-only PDF files.

It is suitable for creating high-compressed PDF files which contain entire documents saved as pictures. The following settings are used:

• Document recognition and synthesis of the logical structure of a document are not performed.
• Skew correction is not performed.
• PDF export is optimized for the minimum size of the resulting file.
• The entire document is saved as a picture (PEM_ImageOnly mode).

XPS export format

Starting from this release FRE supports exporting to XPS format.

This format is developed by Microsoft in contrast to PDF format (see Wiki).

This is optional export format and to enable it one should enable this format in a license.

Retaining document layout in textual export

TextExportParams::RetainLayout property turns on the export mode in which the original layout is simulated by inserting spaces. When displayed with a monospace font, the text and table columns will be level. The distance between blocks will be approximately the same.

This mode is not intended for right-to-left or vertical texts.

When this property is set to TRUE, the InsertEmptyLineBetweenParagraphs property is ignored, and the presence of empty lines is determined by the size of empty space between paragraphs in the original. The ExportParagraphsAsOneLine property is also ignored, and the line breaks are always kept.

This property is FALSE by default.

Distribution is digitally signed

The distribution is signed with “Developer ID Application” certificate. This type of certificate is used to sign a Mac app before distributing it outside the Mac App Store.

Fixed Bugs

There are no bugs that have been reported by Beta testers and fixed in this release.

Mac OS Version Limitations

• DjVu opening
• Scanning
• ICR/OMR
• Visual Components
• WDP/WIC/BITMAP input formats and other Windows-specific functionality
• PDF text layer reusing

Known Issues and Workarounds

The release supports only Open License protection

R1 GM works with Open Licenses only.

Software licenses are planned for R2 release.

Java wrapper is not included into the distribution

Though we have partly functional Java wrapper for the Engine it still misses some important parts.

In some cases current version of the wrapper is enough, so please consult with HQ product analyst in case of urgent need.

We are planning to include fully-functional Java wrapper into R2 release.

No built-in parallel processing

The release does not include built-in multi-page document processing on multiple CPU cores.

The feature is planned to be included into R2 release.

IFontSet::EnablePdfStandardFonts is non-functional

It is not possible to use the standard PDF fonts during synthesis. That may lead to font embedding even though the text is typed using one of the standard PDF fonts.

This will be fixed in the maintenance release.

OnPageProcessed comes once per a document

During opening, synthesis, exporting stages the callback OnPageProcessed arrives only once per a document rather than after each page.

This will be fixed in the maintenance release.

Some API is not implemented

The following API is not implemented in FRE 10 and FRE 11:

• IFootnoteSeries::IsNumberingWithSuperscript. Always returns “false”.
• IFootnoteSeries::PositionOnPage. Always returns “FPPT_SingleColumnSection”.
• IFootnoteSeries::PositionInDocument. Always returns “FPDT_PageEnd”.
• IFootnoteSeries::HasSeparator. Always returns “true”.
• ITextPicture::ColumnNumber. Always returns “0”.
• ICharParams::IsWordStart. Always returns “false”. It is true only for character parameters got through IWordRecognitionVariants interface.
• IIncut::TextWrapping. Always returns “TW_Undefined”.
• IRunningTitlesSeriesText::HasSeparator. Always returns “false”.

The implementation is not planned.

PDF/A validation report

The following issues are known for PDF/A files produced by this release of FRE 11:

1. Adobe Acrobat 11.0.3 reports “Text cannot be mapped to Unicode” for 2% of images in CJK languages recognized and exported into PDF/A-1a or PDF/A-2a formats. On the other hand http://www.pdf-tools.com/pdf/validate-pdfa-online.aspx on-line validator finds no issue in the same documents.
2. callas pdfaPilot, 3.1 (156) and 4 report “Image is not valid” for few images exported into PDF/A-2a (-2u) format. At the same time Adobe Acrobat 10.1.4, 11 report no issues with these files.