OptimiDoc Server Help Center

Processing and OCR

The Processing tab configures three stages that run after capture and before delivery: image enhancement to clean up scanned pages, document separation to split a batch into individual documents, and OCR to add a searchable text layer or convert to an editable format. The OCR engine and the available languages are chosen automatically from the output format selected on the Output Options tab.


https://optimidoc.atlassian.net/wiki/download/attachments/42631246/scan-template-processing.png?api=v2

Screenshot: Scan Workflow editor — Processing tab

How It Works

These steps are applied by the OptimiDoc scan pipeline on the server. First, image enhancements clean each page; then document separation divides the pages into one or more output documents; finally, if the output format runs OCR, the recognition engine produces the searchable or editable result. The OCR section only appears when the chosen output format is an OCR format — image formats (plain PDF, JPEG, TIFF) skip recognition entirely.


Image Enhancements

Each enhancement can be toggled on independently:

Enhancement

Description

Remove Punch Holes

Detects and removes hole-punch marks.

Remove Black Borders

Removes black edges left when a page is smaller than the platen.

Auto Page Orientation

Detects and corrects page orientation from content.

Auto Deskew

Straightens pages fed slightly crooked.

Split Facing Pages

Splits a scan of two facing book pages into separate pages.

Split Dual Pages

Splits a scan containing two side-by-side pages.

Despeckle

Removes small dots and noise artefacts.

Texture Filtering

Removes a coloured background or overlay. Reveals a Filter Color picker and a Filter Type choice (Background or Foreground).


Document Separation

Document separation decides how a batch of scanned pages is divided into individual output documents — essential when scanning several documents in one pass.

Mode

Description

One Document

All pages become a single output file.

By Barcode

A separator barcode starts a new document.

By Blank Page

A blank page separates documents.

One Document + Barcode

All pages form one document, but barcode values are read and can be used in naming and routing.

By Page Count

Documents split at a fixed number of pages.

By OCR Zone

Separation is driven by an OCR zone marked as a separator.

Barcode settings

For the two barcode modes, a Barcode Type can be chosen. The default, All Barcode Types, auto-detects any supported symbology; you can instead restrict detection to one of: Code 39, Interleaved 2 of 5, EAN-13, Code 128, EAN-8, PDF417, Codabar, UPC-E, Industrial 2 of 5, IATA 2 of 5, Matrix 2 of 5, Code 93, PostNet, UCC-128, Patch, Aztec, DataMatrix, QR Code and UPC-A. Two further options are available:

  • Remove Barcode Page — discards the separator page from the output.

  • Barcode Pattern — a regular expression that restricts separation to barcodes whose value matches (for example ^DOC-[0-9]+$).

Page count settings

For By Page Count, set Pages Per Document — the fixed number of pages in each output document.

Zone separation settings

For By OCR Zone, choose the separator zone. The dropdown lists only zones flagged as Separator on the OCR Zones tab; if none is flagged, a prompt asks you to mark one there first. A separator zone is required to save the workflow in this mode, and the editor warns you if the chosen zone is later deleted. An optional Zone Pattern (regex) can further constrain when the zone value triggers a split.


OCR Settings

The OCR Settings section appears only when an OCR output format is selected. OCR is enabled automatically by that choice — there is no separate on/off switch here.

Recognition engine

OptimiDoc includes two OCR engines, and the output format decides which one runs:

  • ABBYY — used for the ABBYY output formats (searchable and compact PDF, Word, Excel, PowerPoint, RTF and Text). Highest accuracy and supports the full language list. Requires the ABBYY OCR licence.

  • OptimiDoc — the built-in engine, used for the OptimiDoc (Od) formats. Always available. Its language list is filtered to the languages the engine supports; if you switch to an OptimiDoc format while unsupported languages are selected, they are dropped and you are notified.

Language

OCR Language is a multi-select — tick one or more of the 40-plus supported languages to match the documents being scanned. Selecting the correct languages significantly improves accuracy. The list shown reflects the engine chosen by the output format.

Recognition Mode

  • Accuracy — favours the best possible recognition, at the cost of speed.

  • Speed — favours faster processing for high-volume scanning.

Blank Page Handling

Option

Description

Keep All Pages

Every page is kept, including blank ones.

Use Device Detection

Relies on the MFP's own blank-page detection.

Use OCR Detection

Lets the OCR engine detect and drop blank pages.


How OCR Runs on the Server

OCR is CPU-intensive, so it runs through a bounded pool rather than all at once. The number of documents recognised in parallel is capped by the server's Scan processing concurrency (the MaxTasks value, configured under Scan Settings); the default is a conservative two parallel tasks, which can be raised on capable hardware. As a safety net, the OptimiDoc engine abandons any single document after a 60-minute timeout and logs a warning, so one problematic scan cannot block the pool indefinitely.


Best Practices

  • Enable Auto Deskew and Remove Black Borders for feeder workflows — the most commonly needed clean-ups.

  • For barcode separation, print separator sheets with a clear, high-contrast barcode in a consistent position.

  • Use Accuracy mode for important documents and Speed mode for bulk scanning.

  • Select every language that appears in your documents — the engines handle multiple languages in one document well.


In Summary

The Processing tab is where a raw scan becomes a clean, correctly-divided, searchable document. Enhancements repair image quality, separation turns a page batch into the right number of documents (by barcode, blank page, page count or OCR zone), and OCR — driven automatically by the output format and its engine — adds the text layer. The under-the-hood pool and per-job timeout keep recognition throughput predictable even under load.


Related articles