Reducing Manual Data Entry For Handprint and Handwriting

Forms have been filled out by hand since there very first form was invented. When the first automated forms processing solutions were invented a few decades ago, they avoided hand-filled forms in favor of typed forms. For the earliest recognition engines, machine typed print was consistent and easier to identify. However, in the 1990s, the recognition vendors took on the challenge of handprint and, using a branch of artificial intelligence called neural network technology, developed new techniques to read hand-printed characters. They called it Intelligent Character Recognition. Many use the term "ICR" to refer exclusively to handwriting recognition.

Good Form Design Can Help Capture Handprint

ICR works best on neat handwriting.  Recognition is best when the characters are separated, touching each other as little as possible.  That's why form design is so important if you are considering using ICR. A form design for handprint capture might include “constraint boxes” – faint squares that encourage form fillers to print within the box, and keep letters separate.  The squares can be made to drop out during scanning, leaving only the hand-printed data.

Free Form Handwriting Results Are Going Up

The recognition engine vendors have not ceased to innovate.  In recent years, there are ICR engines that perform reasonably well on free form, or cursive, handwriting. They use very sophisticated algorithms to break down each letter and leverage statistical probabilities to identify each word.

It's All About Increasing Productivity

Capturing handprint is only as good as the penmanship of the writer. There are some written script that experienced data entry people cannot fathom.  A recognition engine will do no better.  ICR makes a lot of sense when you look at it this way:  If you are able to capture 50% of what you were previously entering manually, you have raised productivity and lowered cost.

Datacap Taskmaster offers a full selection of OCR and ICR engines for recognition of all types of data, including hand print. Datacap’s rules-based configuration tool, Datacap Studio, allows users to apply different recognitions to different fields – or multiple recognition engines – with voting – to the same field, or combinations of both.