New Handwriting Algorithm Replicates Writing

New Handwriting Algorithm Replicates Writing

Although various methods have been tried previously that combine handwriting and typing, none have been able to get completely to grips with it. There are stylus pens that consumers can but to work on tablets, but they are often very unresponsive and don’t do your handwriting much justice either. Other methods include programs that attempt to convert your handwriting into fonts but the scanning process is unreliable and often leaves gaps in the lines where it does not pick up the ink evenly.

But, there may be a light at the end of the tunnel for those of you who want nothing more than to combine your handwriting with typing and it comes in the form of an algorithm that analyzes handwritten text and then copies that exact style with any combination of words. The machine was designed by researchers at University College London and was built around glyphs (specific instances of a character). It has the ability to recognize a person’s character choices, the joining up between letters, vertical, and spacing.

New Handwriting Algorithm Replicates Writing 800 × 390

The project is aptly named, “My Text in Your Handwriting” and has the potential to help those that are unable to hold a pen or if you simply want to send a typed letter to someone but want it to look handwritten. One of the studies authors, Dr. Gabriel Brostow, mentioned that is could also be used in helping to detect forgeries as the handwriting specimen can be examined under a microscope. The software would then characterize the handwriting to predict the odds that something had been forged.

The algorithm was tested out on a number of different willing participants. They were asked to write out familiar, English-language passages from the top 100 books in the Project Gutenberg database. Once completed, people have been invited to distinguish between true handwritten envelopes and those that had been created by the software, and around 40 percent of people guessed incorrectly. So although a 60 percent success rate is good, there is still much room for improvement here, and hopefully, more developments will soon follow.

More News To Read