Content bases recognition – ground breaking technology to boost read rates


The latest development from Prime Vision narrows the gap between human and machine recognition. Called Context Reco, it is an extremely powerful technique that is resulting in even higher read rates for both handwritten and machine printed text. 

To date, Prime Vision systems have been mainly based on High Yield Character Recognition (HYCR®).  It uses special patented features to recognise individual characters and patterns that have been hand written or machine printed.

Context Reco complements HYCR technology and, as its name implies, it looks at features contextually.  This could be the relative position and size of the character in relation to the line or features of neighbouring characters or patterns.

“To decode a particular character Context Reco takes around 50% of the information it needs from the character itself and the remaining 50% from its relationship to adjacent patterns,” explained the systems developer, Prime Vision’s artificial intelligence specialist, Dr. Marius Bulacu.

The new recognition method mimics the way a human would rationalise the character or word, by capturing the information visually whilst applying intelligence to make sense of the received image.  In Context Reco character recognition and contextual relationship are not sequential processes but co-ordinated and mutually beneficial methods that have a positive impact on the reading outcome.

After being tested internally and its impact measured against different language sets from across the world, Context Reco was then applied to specific customer projects and the results were significant.  A 15% reduction in video coding requirement was typically achieved.

Context Reco is now available to posts across the world as a module for both the ParcelMatch and MailMatch open architecture platforms.  And it allows Prime Vision to offer systems that are one step closer to the ultimate goal of 100% readability.

PV122 handwritten_package


Pressemitteilung veröffentlicht am 27.11.2013 in News (In- und Ausland), Software, Sonstige Produkte / Services / Dienstleistungen.