Can Software “Read” (really) Handwritten Documents?

Posted on: June 22, 2021

Yes. Simply stated, computer software called Intelligent Capture can “read” a document and convert the handwritten words into data. In fact, many documents like correspondence, checks, applications, orders, faxes, letters and packages sent via postal mail are read every day by Intelligent Capture software. Interested? Read on…

The Challenge – Documents vs. data

We fill out forms virtually every day. They are mostly annoying, sometimes long, often confusing and worst of all, require us to fill them out with a pen or pencil. I just completed one at my doctor’s office last week. Interestingly, the problem of forms only starts with someone completing the form.

It’s also a challenge for millions of U.S. businesses that receive paper forms and correspondence each day. You may be thinking that most correspondence in 2021 is electronic, sent via email, phone apps or uploaded to a web site. It is true that the majority of correspondence between individuals and companies today is digital.  However, it remains very common for businesses to receive hundreds or thousands of pieces of postal mail, faxes and PDFs every day.

Ever wonder what happens to those forms, correspondence documents and faxes?  How do they turn into data on some computer screen or database? That’s the real challenge. Those documents are analog, not digital. Analog documents are unusable to most systems until they are digitized and converted into data.

Recognition – Images into Characters

Most paper forms are scanned/digitized and sent to software call Optical Character Recognition (OCR).  This software was pioneered in the 1930’s, revolutionized in the 1970’s by Ray Kurzweil and progressed at a modest pace through the 1980’s and 1990’s.

Optical Character Recognition (OCR) software attempts to convert an image of a document, taken from phone, photo, scanner, email or fax, into data. At a high-level, OCR looks at the image of each character of each word on a page and uses algorithms to recognize or “read” the pixels contained in the image as a character in the alphabet. OCR traditionally works with typed, or machine printed characters, check boxes, bar codes, signatures and other types of data, but was typically much less graceful with handwritten information. In other words, most OCR software did a terrible job with most handwritten information. Fortunately, that changed some time ago, but it seems the word never got out.

The Advent of Handwriting Recognition

Sometimes referred to as Intelligence Character Recognition (ICR), handwriting recognition was developed about 25 years back to take on problems of high-volume handwriting recognition like checks, surveys and postal address recognition on envelopes.  The common misunderstanding is that handwriting recognition produces low quality results, but the truth is that handwriting recognition can work as well as OCR in most circumstances, especially when the right recognition engines are combined with Artificial Intelligence and Machine Learning.

The Quantum Leap – AI and Machine Learning spawn Intelligent Capture

About five years ago advances in Artificial Intelligence and Machine Learning (AI) changed the game for OCR and handwriting recognition. AI added the ability for the OCR and handwriting engines to quickly train themselves and get smarter over time.  The combination of AI + OCR + handwriting recognition created a game-changing technology that is commonly referred to as “Intelligent Capture” or “intelligent Document Processing”, while some still refer to the category of technology as OCR.  Instead of expert technical resources writing complex, custom-built computer scripts to setup a recognition solution as in the past, now AI can do most of that work with software and CPUs. Also, in place of technicians constantly tweaking systems as forms change, AI can proactively recommend tweaks that can easily implemented to improve performance and results. Improved results can reduce costs, improve speed and quality by reducing data entry.

Conclusion – Intelligent Capture with AI and Handwriting Recognition can read most documents

My experience over the last several years is that Intelligent Capture with AI really has changed the recognition landscape. Especially in relation to handwriting, our customers are seeing real changes in their business processes. Intelligent Capture can be quickly deployed in weeks versus months. By quickly and efficiently turning documents into data businesses can transform their processes with sometimes dramatically reduce costs.

If you would like to learn a bit more about how Intelligent Capture and how BRYJ might help you with your handwriting or other recognition requirements, shoot me an email at mike@bryjinc.com. Thanks for reading and we welcome your comments.