9.2

Edit

TOD OCR Business Process

The TOD approach is now part of publicly available BP for SPA 9.x series. This TOD process uses standard OCR xml output to split text into words, calculates absolute positioning coordinates for each word and generates HTML where original page is used as background while the html words are put as overlay on top of the image.

Features

The Standard OCR BP:

  • Implements asynchronous model for OCR Rest API calls.
  • Uses a Data Store to cache OCR results. You will have to configure a Data Store name for an OCR process to run successfully. When caching is not necessary, you may skip _ocr_cache variable.
  • Uploads all intermediate OCR-related results to S3 bucket. You should configure the S3 bucket for the successful OCR process run.
  • Provides an opportunity to use either Standard OCR approach or TOD OCR approach.

Prerequisites

  • SPA v9.0+

  • ocr_url and convert_lib_path variables are configured in the Global Variables Data Store

    Example:
    convert_lib_path = convert ocr_url: https:// instance name ocr1.workfusion.com/api/v1/cloud

  • ImageMagick installed to perform multi-page document split (if the TOOD OCR type is used)

  • Input data must contain original_document_url

  • Tagging Over Document option requires the following settings to be set for all documents:
    Export Format =xmlForCorrectedImage,
    CharPosition = true

BP assembling instructions

  1. Import Standard OCR BP suitable for your version from the Artifacts into Control Tower.

  2. Configure OCR parameters via ETL Settings in the OCR Settings step.

Artifacts

VersionArtifact
Version 9.0OCR+2.5.1+v+9.0.zip
Version 9.2OCR+2.5.1.zip

OCR settings

  • Bot Config Name: OCR settings 2.5.1

  • S3 bucket: doc-upload

  • OCR Type: TOOD

  • Document language: Japanese,English. This parameter can contain several language names separated with commas, for example, "English,French,German".

  • Allowed Region Types: BT_AutoAnalysis, BT_Text, BT_VectorPicture

  • Change DPI: 300

  • Switch for CharPosition option: false

  • Discard Color Image: Leaves only black-and-white plane in the prepared image

  • Remove Color Objects Type: Removes color objects from the image, modes values: Background, Full, Stamp

  • If use line height in letter size computation: True

  • Low Resolution Mode: True

  • Remove Garbage Size: Auto

  • cache_datastore: ocr_cache_sec

Barcode recognition is not working with TOOD OCR: page is empty with the Barcode region specified.

Use Standard OCR for this case.

Securing links

All links in OCR BPs can be secured using secure S3 bucket. Note that when generating secure links using s3-put plugin, the expires-in-seconds parameter denotes the time after creation when the link will stay live. If this parameter is used, the OCR BP does not cache any documents.

If expires-in-seconds is empty, secure links are not created.

Output

The Business process contains Manual Tasks for both Standard and TOOD branches.

ocr_document_url can be found in the Results tab of these Manual Tasks, or in the Results tab of the previous Prepare URL step.

← PreviousNext →