'ocr' Service

enaio® • 11.10 »

The 'ocr' service converts image documents into text documents that are used for full-text indexing, together with the text recognition software ABBYY FineReader. It can also be used to create PDF files with hidden text, various PDF/A formats, and highly compressed PDFs.

Configuration

The 'ocr' service works with the following default settings:

PDF profile: format PDF/A1b
PDF profile: method Balanced
Text profile Predefined: TextExport.ini
File transfer to enaio® rendition-plus Stream
Number of cores for ABBYY FineReader 1

These settings can be changed via the ocr-prod.yml configuration file located in the\servicemanager\config\ directory.

Example of a configuration in the ocr-prod.yml file:

finereader:
  profile:
    pdfa: PDFA1bBalanced.ini
    text: TextExport.ini
  engine:
    numberOfCores: 1
rest:
  transferPolicy: stream

The example corresponds to the default settings.

Only the settings that differ from the default settings need to be specified.

  • Customizing the PDF profile

    To customize the PDF format and the method, assign a value to the finereader:profile:pdfa property, whereby the value is composed of the format and method: <format><method>.ini

    The following formats can be created:

    Format Notation
    PDF PDF
    PDF/A1a PDFA1a
    PDF/A1b PDFA1b
    PDF/A2a PDFA2a
    PDF/A2u PDFA2u
    PDF/A3a PDFA3a
    PDF/A3u PDFA3u

    The following methods are available:

    Method Description
    MaxQuality Produces results with the best resolution. Speed and degree of compression are secondary.
    MaxSpeed Produces results according to the fastest method. Quality and degree of compression are secondary.
    MinSize Produces results with the smallest file size. Speed and quality are secondary.
    Balanced Produces results with a balanced ratio of quality, speed, and degree of compression.

  • Customizing the text profile

    Currently, only the text profile TextExport.ini is available for creating texts.

Assign a value to the rest:transferPolicy property to specify how the files are transferred.

Transfer Types:
File transfer Description
stream Transfer via an HTTP stream
fileref Transfer via file system reference
auto

The transfer type is selected automatically. The IP address of the endpoint of enaio® rendition-plus is used to determine whether enaio® rendition-plus and the 'ocr' service are running on the same computer. If they are, the transfer is done via file system references; otherwise, it is done via an HTTP stream.

The maximum number of cores ABBYY FineReader can work with depends on the license that was purchased.

Entry: finereader:engine:numberOfCores: <number>

Integrating a Profile File

You can customize the profile or create your own profile file with additional settings and integrate it via the ocr-prod.yml configuration file.

Integration example:

finereader: 
  profile:   
    pdfa: 'file://d:/enaio/OCRconfig/custom_ocr.ini'
engine:   
  numberOfCores: 4
rest:   
  transferPolicy: 'auto'  

Example of a profile file:

[PDFExportParams]
Scenario = PES_Balanced
PDFAComplianceMode = PCM_Pdfa_1b

[PrepareImageMode]
CorrectSkew = false

[PagePreprocessingParams]
CorrectOrientation=true
CorrectSkew=TSPV_No
CorrectGeometry=TSPV_No

[RecognizerParams]
TextLanguage = German,French,English
DetectLanguage = true
BalancedMode=true

[PageAnalysisParams]
DetectVerticalEuropeanText=true

[ObjectsExtractionParams]
DetectTextOnPictures=true

You can find information on the settings in the ABBYY FineReader documentation.

Examples of settings areas:

[PDFExportParams] Setting the parameters for exporting recognized text into PDF format.
[PagePreprocessingParams]   Setting parameters for preprocessing pages.
[PrepareImageMode] Setting parameters for optimizing images prior to processing.
[RecognizerParams] Setting recognizer parameters such as language settings.
[PageAnalysisParams] Setting parameters for layout analyses
[ObjectsExtractionParams] Setting parameters for extracting objects