Optical Character Recognition (OCR)

<< Click to Display Table of Contents >>

Navigation:  Message Retrieval Accounts > Reading/Receiving Messages >

Optical Character Recognition (OCR)

ThinkAutomation supports the extraction of any detected text within an image.

 

As of the current version, the application requires a connection to Microsoft Azure's Cognitive Services to run the text recognition process. If OCR is enabled in the Account Properties, File Pickup account is used and the file is a supported image type, ThinkAutomation will extract any text detected within the image and add it to the message body where it can be further parsed for use in triggers.

 

Prerequisites for use.

 

1. An active Microsoft Azure subscription.

2. An active Azure Cognitive Services resource.

 

Setting up a Microsoft Azure Account

 

Microsoft has provided guidance on how to quickly and easily create an Azure subscription with an initial free period.

Please follow the guidance outlined here.

 

Creating an Azure Computer Vision API resource.

 

Please follow Microsoft guidance outlined here to create the Azure Cognitive Services resource required.

 

Configuring OCR in ThinkAutomation

 

OCR is enabled at the account level, meaning it is enabled from the Account Properties of the Account in which you wish to extract text from images.

 

1. In Account Properties, check the Enable OCR check box. This will make an OCR Settings button appear.

 

OCRButton

 

2. Click OCR Settings to open the Azure configuration screen.

 

AzureOCRSettingsScreen

 

3. Enter the Endpoint and a Subscription Key for the Azure resource you created. This can be found in the Azure Portal under Resources - Access Keys as outlined here

 

4. Click OK to save your Azure settings

 

5. Create a File Pickup account to pick up your images. (In this example we are reading PNG files).

 

filepickupocr

 

You can see in the example below that if you open a processed message, the extracted text from the image is in the message body. The original image is also attached.

The example also shows the opened original message, demonstrating the parsing quality of an image containing a passage of text. The same functionality could be used for text contained within images of documents such as driving licenses, or even street signs and car registration plates.

 

ocrresult

 

 

 

ThinkAutomation © Parker Software 2016