HOW TO: Use Word Confidence to improve OCR results

#1 Posted : Wednesday, February 5, 2020 5:16:41 PM(UTC)

Marcus Andra

Groups: Registered, Manager, Tech Support, Administrators
Posts: 107

Was thanked: 9 time(s) in 9 post(s)

With the use of the LEAD Engine for OCR, you are provided with powerful engine settings that allow you to address noise and filter what is being recognized.

Three in particular which you should keep in mind are:

Recognition.Words.DiscardLowConfidenceWords set to true by default
Recognition.Words.DiscardLowConfidenceZones set to false by default
Recognition.Words.LowWordConfidence set to 50 by default

Using these settings, you can manipulate the confidence to allow the removal of unwanted words. The documentation for these settings can be found here: https://www.leadtools.co...ead-engine-settings.html

When loading the below test image into the LEADTOOLS OCR demo and recognizing the page, you will see the results do not come out correctly. In particular, the noise present in the image is providing incorrect output when recognized automatically through the use of the AutoZone method . The OCR demo can be accessed after installing the LEADTOOLS SDK and is found at the following default installation location: C:\LEADTOOLS 20\Shortcuts\OCR - MICR - ICR - OMR\.NET Framework Class Libraries\Main OCR Demos\OCR 64-bit Demo.lnk

To correct this through the use of the OCR demo, you can change the Low word confidence value to 80, and set the Discard zones with low confidence option to true.

With these changes, we now get suitable output which eliminates the presence of the unwanted noise.

You can access these settings in your own application by adding the following code after the OCR engine has been start. For more information on how to incorporate the use of confidence in your application you can look here: https://www.leadtools.co...onfidence-reporting.html

Code:

ocrEngine.Startup(null, null, null, null);
IOcrSettingManager settingManager = ocrEngine.SettingManager;
settingManager.SetBooleanValue("Recognition.Words.DiscardLowConfidenceZones", true);
settingManager.SetBooleanValue("Recognition.Words.DiscardLowConfidenceWords", true);
settingManager.SetBooleanValue("Recognition.Words.LowWordConfidence", true);

LEADTOOLS Test image

Marcus Andra
Developer Support Engineer
LEAD Technologies, Inc.


	Try the latest version of LEADTOOLS for free for 60 days by downloading the evaluation: https://www.leadtools.com/downloads Wanna join the discussion? Login to your LEADTOOLS Support account or Register a new forum account.

Notification

HOW TO: Use Word Confidence to improve OCR results