The OCR and barcode recognition technologies implemented in Recognition Server deliver unprecedented accuracy, support various types of text and support the most popular 1D and 2D barcodes. The OCR process has extensive language support. The supported languages include 198 languages including Latin, Cyrillic, Greek, Arabic, Chinese, Japanese, Korean, Vietnamese, Hebrew, Yiddish and Thai. European languages written in Gothic fonts are also supported.

To preserve the original document layout, ABBYY Recognition Server uses Adaptive Document Recognition Technology (ADRT). ADRT significantly improves document layout retention when saving documents to DOC and RTF formats. The logical structure of an entire document is reproduced, including headers, footers, footnotes, page numbers, table of contents linked to document sections and notes to pictures and diagrams.

Support Many Recognition Languages

• 43 main languages with dictionary support: Arabic (Saudi Arabia), Armenian (Eastern), Armenian (Grabar), Armenian (Western), Azeri (Latin), Bashkir, Bulgarian, Catalan, Croatian, Czech, Danish, Dutch, Dutch (Belgian), English, Estonian, Finnish, French, German, German (new spelling), Greek, Hebrew, Hungarian, Indonesian, Italian, Latvian, Lithuanian, Norwegian, Norwegian (Bokmal), Norwegian (Nynorsk), Polish, Portuguese, Portuguese (Brazilian), Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Tatar, Thai, Turkish, Ukrainian, Vietnamese;
• 133 additional languages without dictionary support: Abkhaz, Adyghe, Afrikaans, Agul, Albanian, Altai, Avar, Aymara, Azerbaijani (Cyrillic), Basque, Belarusian, Bemba, Blackfoot, Breton, Bugotu, Buryat, Cebuano, Chamorro, Chechen, Chukchee, Chuvash, Corsican, Crimean Tatar, Crow, Dargwa, Dungan, Eskimo (Cyrillic), Eskimo (Latin), Even, Evenki, Faroese, Fijian, Frisian, Friulian, Gagauz, Galician, Ganda, German (Luxembourg), Guarani, Hani, Hausa, Hawaiian, Icelandic, Indonesian, Ingush, Irish, Jingpo, Kabardian, Kalmyk, Karachay-balkar, Karakalpak, Kasub, Kawa, Kazakh, Khakass, Khanty, Kikuyu, Kirghiz, Kongo, Koryak, Kpelle, Kumyk, Kurdish, Lak, Latin, Lezgi, Luba, Macedonian, Malagasy, Malay (Malaysian), Malinke, Maltese, Mansi, Maori, Mari, Maya, Miao, Minangkabau, Mohawk, Moldavian, Mongol, Mordvin, Nahuatl, Nenets, Nivkh, Nogay, Nyanja, Ojibway, Ossetian, Papiamento, Provencal, Quechua, Rhaeto-Romanic, Romany, Rundi, Russian (Old Spelling), Rwanda, Sami (Lappish) , Samoan, Scottish Gaelic, Selkup, Serbian (Cyrillic), Serbian (Latin), Shona, Sioux (Dakota), Somali, Sorbian, Sotho, Sunda, Swahili, Swazi, Tabasaran, Tagalog, Tahitian, Tajik, Tok Pisin, Tongan, Tswana, Tun, Turkmen, Tuvinian, Udmurt, Uigur (Cyrillic), Uigur (Latin), Uzbek (Cyrillic), Uzbek (Latin), Welsh, Wolof, Xhosa, Yakut, Yiddish, Zapotec, and Zulu;
• 5 East Asian languages: Chinese (Traditional, Simplified), Japanese, Korean and Hangul (Korean);
• 6 languages for recognition of old European documents and Gothic fonts in books printed in 18-20th centuries
  • English,
  • French,
  • German,
  • Italian,
  • Spanish,
  • Latvian;
• 4 artificial languages: Esperanto, Ido, Interlingua, and Occidental;
• 6 programming languages: Basic, C/C++, COBOL, Fortran, Java, and Pascal;
• Simple chemical formulas
• Digits
• 1D Barcodes
  • Check Code 39, Check Interleaved 25, Code 128, Code 39, EAN 13, EAN 8, Interleaved 25, CODABAR (without checksum), UCC Code 128, Code 2 of 5 (Industrial, IATA, Matrix), Code 93, UPC-A, UPC-E, Patch Code and Postnet;
• 2D Barcodes
  • PDF 417, Aztec, Data Matrix, QR Code
• Multiple Text Types
  • Normal, Fax (mode for low-resolution texts), Typewriter, Dot Matrix Printer, OCR-A, OCR-B, MICR (E13B), Gothic

Automatic quality control allows the administrator to set a threshold for recognition accuracy: documents with poor-quality text will not be converted, but rather stored in a separate folder for special treatment.

Verification Station

A client station for proofreading recognition results. Verification can be enabled for all pages or it can be based on the accuracy threshold. Verification permissions management is supported.

Indexing Station

A client station for document indexing and classification.

1. Multi-Export Destinations

ABBYY Recognition Server enables multiple destinations for data and images as well as generation of searchable PDFs.

2. Flexible File Output Formats

• PDF, PDF/A-1a, PDF/A-1b, PDF/-2a, PDF/A-2b, PDF/A-2u
• RTF
• DOC, DOCX
• XLS, XLSX
• TXT, CSV
• HTML
• TIFF
• JPEG, JPEG 2000
• JBIG2
• PNG
• EPUB
• XML, Alto XML
• FineReader internal format (FineReader Engine-compatible)

3. Available Connectors to Enterprise Systems

• Export to Microsoft SharePoint
• IFilter for TIFF files
• Connector to Google Search Appliance

4. Available Customization and Integration Options

• Custom processing parameters defined via XML files (XML Tickets)
• WEB API
• COM API
• Scripting in VBScript and JScript

ABBYY Recognition Server automatically converts large volumes of paper documents or document images into fully searchable electronic text suitable for business processes including archiving, e-discovery, and enterprise search. It enables automated, unattended document processing that can be managed and accessed from within an organization or remotely. Recognition Server can also connect with a variety of back-end systems and third-party applications, integrating via Scripts, XML tickets, a Web-service API or a COM-based API. ABBYY intelligent OCR and PDF conversion technology delivers highly accurate document conversion with recognition of up to 190 languages.

Architecture

ABBYY Recognition Server consists of several components, which can be installed on one or many computers in a LAN. The main components are:

• Server Manager — a central service component, which controls the document processing queue and distributes the tasks among the stations.
• Processing Station — a service that performs recognition and document conversion.
• Scanning Station — a client station for batch scanning and image pre-processing.
• Indexing Station — a client station for document indexing and classification.
• Connector to Google Search Appliance™ (GSA) — a component that allows Google Search Appliance to use ABBYY Recognition Server for extracting content from document images .
• Connector to Microsoft® Search Systems (IFilter) — a component that allows Microsoft Office SharePoint Server and Windows Search to use ABBYY Recognition Server for extracting content from document images.
• Remote Administration Console — a client console used for configuring and monitoring Recognition Server.

Document Processing

ABBYY Recognition Server processes each image file according to a workflow — a set of processing parameters predefined by the administrator. ABBYY Recognition Server can run several workflows with different parameters simultaneously. Each workflow corresponds to a unique input source (a folder, a SharePoint library or a mailbox).

Processing Steps

A workflow in ABBYY Recognition Server typically includes up to six configurable stages. Each workflow runs independently of others according to its own schedule and priority.

Six Stages of Document Processing

1. Scanning/Import of images. Images can be either scanned by an operator on the Scanning Station and then sent to ABBYY Recognition Server, or automatically imported by ABBYY Recognition Server from an input folder (network folder, FTP folder, SharePoint® library, or mailbox). ABBYY Recognition Server arranges image files in a queue to process them automatically according to priorities.

2. Recognition. The OCR process runs automatically on the Processing Station. If several Processing Stations are installed in the system, the files will be distributed among these Processing Stations evenly for optimal performance. Deploying additional Processing Stations brings a linear increase in OCR speed.

3. Verification (optional). In some cases, for example when digitizing books, verification of the recognition results might be necessary. Verification Stations allow operators to check all documents or only documents below a certain accuracy threshold.

4. Document separation (optional). When the batch scanning or import is performed, a document separation may be required. The documents can be separated using blank separator sheets, barcodes or by fixed number of pages per document. Separation can also be done according to a scripted rule.

5. Classification and indexing (optional). Indexing of documents can be done either automatically by a script, or by an operator on the Indexing Station, which allows the operator to manually select the document type and assign document attributes. The operator can also verify the data that has been populated by the script.

6. Export. In the final stage, ABBYY Recognition Server delivers the output documents to their destination (which can be a network folder, a SharePoint document library, or an e-mail address). Additionally, scripts can be applied for intelligent routing and delivery of documents to ECM systems based on document types and attributes.

With the priority management and scheduling features, the administrator can control the order in which the documents are processed and use the stations’ hardware resources efficiently by scheduling OCR for night hours or weekends.