How do I get the best ICR accuracy?

Intelligent Character Recognition and Handwriting

So, how do you get the most out of your ICR Software?  How do you get the most accurate results?  See below:

1. Properly built forms.  Spacing is one of the key requirements in Intelligent Character Recognition.  Forms with boxes or combs will result in the highest accuracy reads.
2. Character Filtering Sets.  Being able to designate fields as numbers or text will help the ICR engine in determining how it should interpret symbols.
3. Validation process.  If you can, having external validation will help to insure proper reads.

What is Intelligent Character Recognition (ICR) Software?

When scanning and capturing images, we typically want to collect information from the scanned documents.  Below are the different recognition technologies that provide that capability:

Optical Character Recognition - provides the ability to recognize text.

Optical Mark Recognition - the ability to read marks or colored shapes.

Intelligent Character Recognition - the ability to read handwritten characters.

Of the three recognition types (OMR, OCR, ICR), Intelligent Character Recognition is the most difficult, and most problematic.  It typically takes training on one specific subject's writing to be accurate with handwriting, and to read handpriting, typically it requires characters bound by boxes, or "combed fields".

Below are some links to definitions:

Intelligent Character Recognition

Handwriting Recognition