By: Ray Gerwig, Hyland’s senior product evangelist
In the world of document capture and our march toward digital transformation, optical character recognition (OCR) maintains a role of ever-increasing importance. Because documents contain much of the content that runs our business processes, it’s important to identify, qualify and control that content as quickly as possible.
OCR and document capture
In order to speed up and improve content and data ingestion, your organization can use OCR software to automatically identify documents. Perhaps even more importantly, it can also use OCR to extract the content you need from those documents for critical business processes.
It’s no wonder we are all looking for, or already using, tools that speed these documents and content into our business solutions more quickly and with reduced manual data entry. The advance of OCR accuracy has made it dramatically more valuable for business transaction processing since its beginnings, more than 30 years ago.
OCR and image quality
Admittedly, much of this gain is the result of better quality imaging that is available today. We can say, genuinely, that higher quality images provide better OCR results. Image quality is even more important than the OCR engine you use.
This is why I never provide OCR accuracy projections without seeing sample images; and you should be leery of anybody who promises results without seeing your sample documents. OCR capability is still completely dependent on image quality.
OCR for automation and data entry
No matter how good systems get at performing automated document classification and data extraction, there will always be times when we need users to perform verification or manual data entry. But even during these manual tasks, we can provide assistance using OCR.
During document verification steps and manual data entry, OnBase provides OCR assistance to make users more efficient. As they start typing a value, Interactive Data Capture can complete it for them. For a user doing manual data entry, it works just like those familiar search engines, completing values for them and reducing key strokes. They can also just click on the screen while viewing a value, or swipe through a phrase, to have that data automatically entered. It even highlights the value and indicates the OCR confidence level, alerting users when values might require more scrutiny.
The best part is that this functionality is server-based, so you can deploy it quickly and easily throughout your organization without needing to install OCR software on every workstation. We’ve all typed (and probably mistyped) long account numbers or identification codes, so it’s easy to recognize the value of this capability.
It just makes sense to invest wisely and use OCR tools and automation wherever you can during content capture processes, to both save time and improve accuracy. After all, content needs to be accurate and locatable in order to be useful.