9.0

Edit

Upgrade OCR server from 8.2.x to 8.5.x+

License

Repeat the below actions for all activated licenses for all OCR servers.

  1. Run License Manager using any user.

    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<INSTALL_DIR>/apps/ABBYY/FREngine11/Bin
<INSTALL_DIR>/apps/ABBYY/FREngine11/Bin/LicenseManager.Console

    License serial number and other license parameters can differ from samples.

  2. Select activated license:

  3. Open the license parameters:

  4. Make a screenshot of the first page. Ensure that the line with Volume > Regular Text > Remains is visible.

  5. Request and activate the license.

API

Pattern usage

Changes below are required in case custom pattern is used in BP.

See the full OCR API reference.

  1. Replace the patternName parameter with the pattern parameter that is passed by content, not filename.

    Affected endpoints are as follows:

    • /processImage
    • /submitPattern: added
    • /trainPattern: renamed to mergePattern that is combined with a new one
    • /processDocument: uses pattern submitted with /submitPattern

  2. Use useDefaultPattern=true instead of patternName=default.

    Affected endpoints are as follows:

    • /processImage
    • /processDocument

OCR parameters

  1. Collect all changed parameters of OCR service in YML files (the /etc directory) comparing with standard listed in blocks below:

    ocr-rest.yml:

    standard OCR rest properties

    db.cleanup.abbyy:
input: 2000
output: 2000
record: 600

spring.profiles.active: disabled-auth,mongoDbQueue,gridfs-storage,mongo-task 

ocr.rest.timeout: 5
ocr.rest.retries: 2

    ocr-worker.yml:

    standard OCR worker properties

    worker.executor.abbyy.pool.size: 

spring.profiles.active: abbyy,mongoDbQueue,gridfs-storage

  2. Update configuration YML files with collected parameter values on the previous step.

    Files to review in case any customization exists:

    • ocr-rest.yml
    • ocr-worker.yml

OCR GhostScript

A newer version of GhostScript has impact on document convertion to TIFF and may change quality of OCR.

If one of pre-processing steps uses GhostScript before OCR and quality is worse, there is a need to downgrade version of GhostScript in Bot Task (PDF to TIFF conversion) of Business Process:

gs is a new version, gs919 is the old version.

← PreviousNext →