ChronoScan can read documents with several OCR engines. The engine used for the full-page OCR is selected in the OCR Read menu of the Scan/Input or Data Entry tab (Active OCR engine), and can also be set only for the current job (Use OCR configuration only for the current job). In addition, every OCR Zone or grid column with Advanced OCR Reading enabled can use a different engine from the global one.
Available engines:
ChronoScan OCR (native): included in every installation, multi-language, fully local, no per-page cost. Good results on clean printed documents and native PDF files.
Tesseract (free): open-source engine integrated in ChronoScan; a local alternative at no cost.
Nuance OCR (optional plug-in, sold separately): professional engine that usually gives the best accuracy on scanned documents, produces compressed (MRC) searchable PDF files and, with the ICR option, recognizes handwriting (Nuance ICR Engine). It is installed with its own installer; use the 32-bit or 64-bit plug-in that matches your ChronoScan installation.
Google Vision OCR (cloud): very robust with photos, poor quality scans, irregular print and complex languages. Pay per use with your own Google API key or with ChronoScan credits; Internet connection required. See Google Cloud Integration.
docTR OCR and yodaOCR: additional engines available from the Active OCR engine menu. yodaOCR connects to a yodaOCR server whose address and API key are set in the Program options.
PDF text layer: when a PDF file was generated digitally and contains a text layer, ChronoScan uses that text directly (100% accurate, no OCR cost). It is selected automatically when such a PDF is imported.
Dual OCR: reads the zone with two engines and compares the results. Only what both engines agree on is accepted; any difference marks the field for review. Use it on critical fields such as amounts and account numbers.
Mark Detection: does not read text; it detects whether a mark is present in the zone, for example whether a checkbox is ticked or a signature box has been signed. Used for forms, OMR questionnaires and signature presence. See Checkbox Detection with Data Grids.
Google Document AI (cloud): not an OCR engine as such, but a pre-trained invoice model that returns the invoice fields directly. See Google Document AI for invoice processing.
Recommendation: ChronoScan OCR or Nuance for clean scanned documents, the PDF text layer for native PDF files, and Google Vision OCR for phone photos or poor quality images.
The callouts below describe the engine choices of the Advanced OCR Reading window.
ChronoScan OCR Engine
The ChronoScan OCR Engine is built into every ChronoScan installation and will provide good results whether you have scanned or native PDF documents.
Tesseract is a free, open-source OCR engine (originally developed by HP and later maintained by Google) that can be used as a local alternative to ChronoScan OCR or Nuance OCR.
The Nuance OCR engine is sold separately as the Nuance OCR Plug-In and is recommended for scanned documents, where it usually gives the best results. Install the plug-in that matches your ChronoScan installation (32-bit or 64-bit). A time-limited trial license is available to compare it with ChronoScan OCR and Tesseract.
The PDF-Text Layer is the built in text layer existing in native PDF files. If the PDF files were generate natively (inside any word processing application) this should be the best option to choose from. Whenever ChronoScan detects a native PDF file that contains a PDF-Text layer it will set the PDF-Text Layer as the default Engine to capture text.
The Mark Detection engine does not read text: it detects whether a mark is present in the zone, for example a ticked checkbox or a signature in a signature box. Learn more about checkbox and checkmark detection here.
The Dual OCR engine reads the zone with two OCR engines and compares the results: only what both engines agree on is accepted as the reading, and any difference marks the field for review.
