OCR technology update

To implement the neural network approaches in OCR technologies, ABBYY FineReader Engine was enhanced by the new features of processing the Latin symbols:

• New language model using both for a consistent choice of word variants generated by OCR and for the substitution of new word variants
• End-to-end recognition for Latin-based languages

Machine learning barcode recognition technology

Neural network architecture introduces a new barcode recognition model based on parsing image pixels into two categories: barcode and non-barcode. Regions built around the connected components are further considered as a barcode hypothesis. The recognition process is started using data from each region containing the type of barcode that is highlighted as the most probable.

New technology depends on image resolution and has no dependency on the number of barcode types to detect. This leads to slower processing speed compared to the legacy technology on images of less than 300 dpi resolution and when a single or few barcode types are to be detected.

Improved table structure analysis

With the improved mechanism of document conversion, ABBYY FineReader Engine can detect false vertical separators and correctly process the tables with columns of numbers in the ‘Accounting’ format when a currency symbol ('$') is aligned to the left in all cells.

New ‘Accurate’ recognition mode

The new ‘Accurate’ mode allows you to get the maximum quality of the output document, assuming a reasonable slowdown in the recognition speed. To obtain the best results in processing, use this mode on poor-quality documents and images:

• Invoices and contracts (scans, small text, photographs)
• Receipts (poor print quality of the original receipt, scan, or photo)
• ID documents (text is printed on a complex background with textures or illustrations, holography on the text)

OCR improvements for text near stamps and signatures

To improve the results of recognition for the agreements, a new neural network model for detecting stamps, logos, and signatures is now applied. This model allows the detection of the additional elements in the footers and requisites area of the document, excludes them from the analysis, and highlights the text in the image, ignoring the details. The recognized blocks are superimposed on the image in such a way as if it were a similar document without extraneous marks and stamps.

Better PDF file processing

ABBYY FineReader Engine can recognize various PDFs from image-only (scans) to digitally-born (text and pictures) and tries to reuse all information from them. But there are occasions when PDF files are a mix of image-only and digitally-born pages in one PDF file, thus making them difficult to process. To cover such cases and let customers recognize such files in manual mode, ABBYY FineReader Engine provides new options:

• Adaptive recognition to improve and speed up PDF processing (default PDF recognition mode: PullXTextAndRecognizeRest)
• Text layer quality classifier for preserving good one in the output format (the CheckTextLayer method of the IFRDocument object)
• New content reuse mode for processing PDF or Office documents with mixed content (CRM_ContentAndPictures in SourceContentReuseModeEnum)

Additionally, ABBYY FineReader Engine was enhanced by the option to detect the presence of a digital signature on a document page or inside the PDF (IFRDocument::SourceHasDigitalSignature, IFRPage::SourceHasDigitalSignature).

To ensure more flexible forming the PDF contents, ABBYY FineReader Engine offers the new options:

• Opening PDF Portfolios and processing their contents
• Adding custom images to the output PDF and managing their positions

NeoML (open-source ML C++ library) usage

NeoML is an end-to-end machine learning framework that allows you to build, train, and deploy ML models. This framework is used by ABBYY engineers for computer vision and natural language processing tasks, including image preprocessing, classification, document layout analysis, OCR, and data extraction from structured and unstructured documents.

Key features:

• Neural networks with support for over 100 layer types
• Traditional machine learning: 20+ algorithms (classification, regression, clustering, etc.)
• CPU and GPU support, fast inference
• ONNX support
• Languages: C++, Java, Objective-C
• Cross-platform: the same code can be run on Windows, Linux, macOS, iOS, and Android

Embedded PDFium for processing PDFs

PDFium is a cross-platform native library conforming to PDF standards and controlling all operations related to PDF, including processing, parsing, rendering, and obtaining the output.

Bangla OCR: New supported language

Bangla language has been added to ABBYY FineReader Engine as a technical preview and can now be applied in scenarios with simple document layout preservation.

Arabic OCR

ABBYY FineReader Engine got the new neural network technologies for significantly increasing the Arabic recognition accuracy and correcting the misrecognition of European insertions into Arabic text, losing text strings, and reducing the excessive calls.

Japanese OCR: support of single '℃' Unicode character

To improve the detection of the symbols in Japanese, the output document is corrected by the OCR technologies, so the single '℃' Unicode character goes out instead of the two separate characters '°'.

PowerPoint export improvements

ABBYY FineReader Engine now has a better conversion for the presentation formats, including enhanced layout preservation and the generating of the correct appearance for the output:

• Frames preservation around text elements, paragraphs
• Separators preservation
• Logical grouping of text elements (e.g., text columns)
• Text fitting into a shape
• Preservation of indentation between paragraphs

Enhanced extraction of PDF attachments

Now, you can obtain the list of attachments from PDF Portfolio in the order established when creating a PDF file. Use the IFRDocument::PDFAttachments property to access the collection of the documents extracted from the input PDF files.

Using grace periods for licenses

With the new option, customers can use the ABBYY FineReader Engine license for some time after the expiration date, thereby enlarging the license validity period.

New option for licenses with expiration date

If a license is valid and the expiration date hasn't been reached yet, the 'Expiration Date' field in License Manager contains the exact date value (Expires on %your_license_expiration_date%).

After passing the expiration date, the grace period starts, during which the license is still valid. The duration of this period depends on a particular license (it displays in the 'Expiration Date' field for the license and in License Parameters). Upon the expiration of the grace period, the license becomes permanently invalid

New library to work with PDFs

Now, all processes connected to PDF processing go through the Google PDFium Library.

Export to XLS/XPS

The API required for document export to XLS/XPS format (XLSExportParams, XPSExportParams, FileExportFormatEnum (FEF_XLS, FEF_XPS), IXLExportParams::XLFileFormat, XLFileFormatEnum, XPSExportModeEnum) is now fully available in Linux version.

API Changes

• Changes in recognition mode
  • IRecognizerParams object now has the new ‘Mode’ property instead of the ‘FastMode’ and ‘BalancedMode’ properties. The ‘Mode’ property allows specifying the recognition mode to be applied to the target documents - Fast, Normal, or Accurate mode, each of which provides its own accuracy and recognition speed.
  • RecognitionModeEnum denotes the recognition modes used during document or image processing.
  • The profiles responsible for the speed results have been updated by the enhanced recognition mode.
  • Recognition mode is no longer protected by the license modules, as it was in the previous releases.
  • The CLI sample got the new -rm key for specifying the recognition mode in the command-line-based applications.
• Engine API
  • New IsPdfPortfolio and IsPdfPortfolioFromStream methods to find out if an input PDF is a PDF Portfolio
  • The FREngineDataFolder property of the InitializeEngine function and method now has new default values. Now, the auxiliary engine data and license data are written into different folders that may be re-defined by a user.
  • IsPdfWithTextualContent and IsPdfWithTextualContentFromStream methods working with input PDF text layer are now available in the Mac version.
• Barcode API
  • EnableBarcodesCheck option of the BarcodeParams object specifies whether to use the classifier of the presence of a barcode on an image. Use this property for all barcodes except postal ones.
  • BarcodeTypeEnum got the new BT_AutodetectWithoutPostal constant for detecting the type of barcode automatically, excluding the postal ones.
  • BarcodeTypeEnum got the new BT_JapanPost constant which supports the Japanese Post 4-state Customer Code type.
• Document-related API
  • IFRDocument:
    • ConvertFromOldVersion – this method loads the document saved only by the previously supported versions of ABBYY FineReader Engine, and FREngineVersionEnum denotes these previously supported versions.
    • SourceHasDigitalSignature – this property indicates the presence of the digital signature inside one of the documents.
    • CheckTextLayer – this method detects a text on the specified pages or checks the reliability of a text, e.g., the absence of the broken encoding.
    • AllocatedSize – this property returns the size of the memory allocated for the document in bytes.
    • PDFAttachments – this property accesses the collection of the documents extracted from the input PDF files.
    • New AddImageFileFromAttachments method of the FRDocument object to open an image file from an attachment and add it to a document.
  • IFRPage:
    • SourceHasDigitalSignature – this property indicates the presence of the digital signature on at least one of the pages of the source document.
    • SourceFilePageIndex – this property returns the page index in the source document.
  • Other changes:
    • PDFAttachment object was enhanced by the new ‘FileFormat’ property to define the image format after its opening in ABBYY FineReader Engine.
    • PDFAttachmentBindingEnum now has the PAB_Portfolio constant for PDF Portfolios.
    • FontEmbeddingModeEnum got the new FEM_EmbedFullWhenNeeded and FEM_EmbedSubsetWhenNeeded constants to choose whether the whole font will be embedded into the output PDF or only the subset of a font.
    • FNF_PDF constant of FontNamesFilterEnum is now available in the Mac version.
• Image-related API
  • ImageFileFormatEnum got the new constants to define the image format when using the FileFormat of the PDFAttachments object: IFF_Bmp, IFF_Dcx, IFF_DjVu, IFF_Gif, IFF_Jpeg, IFF_Jpeg2k, IFF_Pcx, IFF_Png, IFF_Tiff
  • The isInMemory property of the ImageDocument object is now available in the Mac version. This property specifies if the image document is stored in memory only or if it is also represented as a folder on disk.
• Text-related API
  • IParagraph object:
    • UserBookmark – this property provides access to a user bookmark by its index in the internal collection of the paragraph's bookmarks.
    • UserBookmarkCount – this property returns the number of user bookmarks in a paragraph.
  • Other changes:
    • The default value of the ‘Spacing’ property of the CharParams and FontStyle objects is now 0.

BCR synthesis

BusinessCardSynthesisParams – this interface is used for fine-tuning business card synthesis. This release allows you to specify the languages that will be used for processing the recognized text. It is also used as the ‘BusinessCardSynthesisParams’ property in the SynthesisParamsForPage object.

SynthesizeBusinessCardEx– this method tries to detect the business card fields in the recognition area using the specified synthesis parameters.

Parameter Objects

TextLayerInjectionParams object has got the new AllowChangePDFAView property for changing the appearance of the output PDF file when facing problems during its processing. It is FALSE by default, so whenever the document is invalid, an error will be returned, and the document processing will be canceled. If it is set to TRUE, a warning about changing the appearance of the output will be displayed.

SourceContentReuseModeEnum got the new CRM_ContentAndPictures constant which is suitable for PDF files and Office documents with mixed content (text and pictures).

The enableExhaustiveAnalysisMode option of the PageAnalysisParams object has been marked deprecated and scheduled for deletion in future versions.

New versions of ALTO (4.0; 4.1; 4.2) are now supported (see AltoExportParams and AltoVersionEnum).

The detectSpacing property of the FontFormattingDetectionParams is now FALSE by default.

New methods for attaching the user-defined pictures to the output PDF:

• PDFPicture interface representing a picture itself and the methods to manipulate the picture position.
• PDFPictures interface representing a collection of the PDFPicture interfaces.
• New ‘PDFPictures’ property of the PDFExportParams object to access the PDFPictures interface.

License-related API

The VolumeRefreshingDate method of the ILicense object obtains the renewal date of a license with limitations on the number of pages processed.

VRP_AbsoluteMonth and VRP_AbsoluteYear constants of VolumeRefreshingPeriodEnum support the refreshment of the counter on a specific date.