Structured Forms for OCR


What are structured forms?

Structured forms are documents that are divided into data fields according to a uniform structure for automated processing. With structured forms, hundreds of individual documents can all follow a single format, making them easier to process by machine. Different data fields contain spaces for different pieces of information. When the documents are scanned, the data fields are identified, text is identified using character recognition software, and information is extracted into a database. Structured forms eliminate manual data entry, reduce errors, allow for easy information validation, and simplify calculations.


How to set up structured forms for OCR

There are some important rules to follow when creating structured forms for automatic processing. These considerations help ensure the accuracy of automatic processing, reducing errors. One important thing to pay attention to is the spacing of the data fields: it is essential for the fields to be spread out sufficiently to avoid confusion of text belonging to different fields. Additionally, depending on whether your forms processing program can recognize handwritten text, you should restrict the contents to printed information, or allow yourself the option of including spaces for handwritten responses. If your software has ICR in addition to OCR, then you can incorporate handwritten text into the forms that you process.


Understanding OCR for structured forms

You should understand how OCR works in order to process structured forms as efficiently as possible. For example, knowing that OCR relies on precise shapes explains why it is important to minimize stray marks on a page, because this can make one text character look like a completely different character. Using OCR for structured forms can help you automate the processing of many documents that you might now even have considered as forms, such as invoices and financial records. However, once you begin using OCR you will quickly see how useful the software can be.



[ Back ]