The OmniPage Capture SDK 16 includes multiple recognition modules, which can be applied on a per-zone basis. This allows a wide range of different types of textual and non-textual data to be recognized, even if they all appear on the same page. For each zone, the application can assign any module available in the configuration or the choice of module can be set as automatic. All recognition modules are under the same API, so development of applications with multi-type recognitions are made easy.

The OmniPage Capture SDK delivers unmatched flexibility and the most accurate results possible by providing multiple machine print OCR recognition engines that can be used individually or in combination depending on individual application accuracy and speed requirements.

The OmniPage Capture SDK's scalable voting interface provides developers with 2 pre-made voting interfaces and direct access to 3 leading OCR engines.

Voting Engines

Speed development and achieve market-leading accuracy by using pre-made OmniPage or PLUS voting interfaces:

• OmniPage (3-Way Voting)
  Achieve the unmatched accuracy level of the world's leading OCR application, OmniPage Professional 16, by integrating the OmniPage Capture SDK's powerful OmniPage 3-way voting engine.
• PLUS (2-Way Voting)
  Integrate the OmniPage Capture SDK's PLUS 2-way voting engine for accurate and fast machine print OCR capabilities.

Individual Machine Print OCR Engines

Integrate the Nuance MOR, MTX or Fireworx OCR engines to achieve optimal results for your specific application:

• MOR OCR Engine
  Supports 119 languages
  Supports a maximum of 500 zones on one image
  Supports Omnifont, Draftdot24 and OCR-A filling methods
  Supports character training to achieve improved accuracy
  Provides 3 page-level accuracy and speed trade off settings including Accurate, Balanced and Fast
  Provides Checking Subsystem based correction
• MTX (Mtext) OCR Engine
  Fast selectable OCR engine
  Support for 12 languages
  Supports a maximum of 64 zones on one imagev Supports Omnifont, Draftdot9 and Draftdot24 filling methods
  Provides 2 page-level accuracy and speed trade off settings including a combined Accurate & Balanced value and Fast
  Provides Checking Subsystem based correction
• FireWorX OCR Engine
  Optimized for speed
  Support for 54 languages
  Supports a maximum of 2,500 zones on one image
  Supports Omnifont filling methods
  Supports character training to achieve improved accuracy

See Throughput Management for more information about the accuracy and speed of these engines.

Developers have a choice of integrating 2 ICR (handprint OCR) recognition modules including a numbers only module and an alphanumeric handprint recognition module.

Numeric Only Module designed for a restricted character set. This recognition module can recognize the following:

• digits (0-9)
• plus sign (+)
• minus sign (-)
• period or full-stop (.)
• comma (,)

Typical recognition time is 280-310 characters per second (Measured on a computer with 1500 MHz Pentium processor / 256 MB RAM running Windows 2000.)

The Alphanumeric ICR Module recognizes hand-printed characters of both upper and lower-case handprint of nearly 100 languages, 15 with dictionary support including:

• Catalan
• Czech
• Danish
• Dutch
• English
• Finnish
• French
• German
• Hungarian
• Italian
• Norwegian
• Polish
• Portuguese
• Spanish
• Swedish

Typical recognition time is 150-220 characters per second (Measured on a computer with 1500 MHz Pentium processor / 256 MB RAM running Windows 2000.)

Barcode recognition modules are available for the recognition of both 1D and 2D barcode types, in black and white, color, or grayscale:

Supported 1D Barcode Types

• Code 3 of 9
• Code 2 of 5
• Code Codabar
• Code I25
• Code 93
• Code 128
• Code UPC-A
• Code UPC-E
• Code UCC 128
• Code EAN 8
• Code EAN 13
• Code Postnet (Postal)
• Code Patch Code
• Code Airline 2 of 5
• Code Planet
• Code BCC 32
• Code 3 of 9 NSS

Supported 2D Barcode Types

• Bar Code PDF-417
• Data Matrix

The OMR recognition module is used for recognizing optical marks. Typical areas of application include questionnaires, educational tests and in reporting or ordering sheets, where the documents being processed are form-like and filled in by respondents.

The frame can be a rectangle, a hexagon, a circle, or an ellipse and can be filled in by any method (x, tick, hatching, horizontal or vertical lines, etc.). The frame may be visible or invisible (scanner dropout) in the image sent for recognition.

Recognition accuracy is 99.95 % and the average recognition time is 0.005 seconds per checkbox zone (measured on a computer with 166 MHz Pentium Processor / 16 MB RAM).