Libraries

Optical Character Recognition (OCR) - Getting Started: Other Tools

Basic overview of several tools (both open source such as Tesseract and commercial such as Adobe Acrobat) that perform optical character recognition (OCR).

Other Tools

There are a number of other tools that provide OCR capabilities. Some, such as ABBYY FineReader and OmniPage are commercially available. Other programs such as Evernote and OneNote are primarily intended for other purposes, but contain OCR capabilities.

The Source conducted tests on several open source, cloud-based, and commercial OCR options and describe their results in "Our Search for the Best OCR Tool, and What We Found."

Transkribus

Transkribus is an open tool for transcribing historical documents, and is one of the few tools detailed in this guide that is geared towards transcribing handwritten text. Transkribus is available for both Mac and Windows.

There is a How-To guide and a Wiki that details how to install and use Transkribus.

ABBYY FineReader

Tool Overview

ABBYY FineReader is a commercial tool available in both Basic and Corporate versions. ABBYY works on both Mac and Windows. The Corporate version has more output management features, document comparison features, and automated batch processing capabilities. The version used in this test is ABBYY FineReader 10 Professional, an older version from 2010/2011. 

Capabilities (Basic)

  • Multiple inputs: Scanner, Camera, image file (TIFF/JPG tested), PDF/PDF/A

  • Multiple Outputs: plain text, Microsoft Word/Excel/PowerPoint, Adobe searchable PDF/PDF/A, image

Screenshot of common tasks in ABBYY FineReader

  • Multi-Language (15+) capabilities, but must be careful which is selected. Incorrect language leads to poor text recognition. Non-English languages can be less optimal in the QuickTask conversions.

  • Collaborative features in output (comments, security, redaction) (Corporate version only) 

  • “Hot Folder” for scheduling of batch OCR (Corporate version only) 

  • Documents are divided into blocks by proximity and spacing.  Can define areas of document as different types (text, picture, table, etc.)  

  • Can include/exclude pictures in output. 

  • Can include entire original image and recognized text in output 

  • Comes with a good selection of workflows; can easily create others to use later 

  • OCR can be trained to improve accuracy, but good results can be obtained with default settings.  

  • User interface makes it easy to see original, scanned document and correction area all at once.  Settings for languages and output formats are easily selected; advanced settings are available through the menus. 

screenshot of workspace in ABBYY FineReader

Notes:  Some samples didn't OCR well using the QuickTask function; all did better when OCR'd inside the app, with choices other than the defaults.  The user interface makes it easy to customize OCR settings. 

OmniPage

Tool Overview

OmniPage is a commercial software made by Nuance, and is only available for Windows. It is available in Standard and Ultimate versions. This test was conducted with a Standard version released in 2014 or before. 

 

Capabilities

 

PDF editing features 

Multiple inputs:  Scanner, Camera, image file (tiff/jpg tested), PDF/PDF/A 

Multi-Language capabilities, but must be careful which language is selected 

Ultimate version does batch OCR 

Can define areas of document as different types (text, picture, table, etc.)  

Can include/exclude pictures in output 

Comes with a good selection of workflows; can create others to use later 

OCR can be trained to improve accuracy 

 

I was testing conversion to Word, but actually converted to .rtf and then opened files in Word. 

 

TIF files output to Word: Files had to be renamed from TIFF to TIF before conversion would work. ku_broadsides_206_OBJ.tiff would not convert at all. 

 

JPG files output to Word: The four ku-… files will not convert. They remain as images even though the process appears to convert to Word. All of the survey files converted correctly. Attempted to rename ku- files thinking that might be the problem, but nothing worked. 

Google Drive

Tool Overview

Google Docs/Google Drive provides the capability for quick and dirty conversion but with a lot of limitations. Text recognition from PDF and JPG of the same image resulted in very different output. 

Capabilities

Prepare the file. These tips will give you the best results: 

Format: You can convert .JPEG, .PNG, .GIF, or PDF (multipage documents) files. 

File size: The file should be 2 MB or less. 

Resolution: Text should be at least 10 pixels high. 

Orientation: Documents must be right-side up. If your image is facing the wrong way, rotate it before uploading it to Google Drive. 

Languages: Google Drive will detect the language of the document. 

Font and character set: For best results, use common fonts such as Arial or Times New Roman. 

Image quality: Sharp images with even lighting and clear contrasts work best. 

Convert an image file with Google Drive/Docs

  1. Load files to Google Drive 
  2. Right click on file, select Open With....Google Docs
  3. The file will be converted to a Google Doc, with the original image at the top and the editable, OCR'd text below.  
    • Bold, italics, font size, font type, and line breaks are most likely to be retained. 
    • Lists, tables, columns, footnotes, and endnotes are likely not be detected. 

There is an Google Drive API which could (probably) be used to do batch conversions from the command line.  

Google Cloud (paid service) offers some more sophisticated computer vision tools including OCR 

 

Tools with OCR Capabilities

Evernote

Evernote can do OCR for searching of images within notes, but does not extract the text or give you editable text—it is only used for searching text within collections of images. Can be useful when organizing files and images. (I think this is only available with the paid version of Evernote 

 

OneNote

OneNote is a commercial product from Microsoft. The desktop version is a free download for Windows and macOS; free mobile apps are also available. The cloud version, available through Office 365, is available to KU faculty, staff, and students via KU MyCommunity. The OCR capability only works in the desktop version of OneNote, not the web version (though results sync with the web version in real time). Once OneNote is open, click Insert and navigate to a saved file. Insert a printout of that file, then just right click and 'Copy text from image.'  The results are much better with the TIFF files than with the JPEG files.