Recognition of Hand-Printed Characters on Mobile Devices

The usage of mobile devices has increased dramatically in recent years. These devices allow a variety of user input such as touch-screen hand gestures, motion, camera, and audio recording. More recently, many apps are allowing the user to print characters using their fingers, followed by the identification and recognition of the characters the user is printing. There are many challenges presented when implementing algorithms performing this character recognition - a small screen along with limited dexterity by using only one finger - results in a less than optimal letter drawn by the user. This paper explores the challenges of performing this character identification and presents a novel scale/rotational invariant algorithm applied to the recognition of these hand-drawn letters. The results of the algorithm are compared to those obtained using a popular Optical Character Recognition (OCR) application, Tesseract, that is often integrated with iPhone Apps for this purpose. The applications for this research are many, but one primary area of interest to the author is an evaluation test for finger dexterity of Occupational Therapy patients. The algorithm is implemented as a fully functional app developed for Apple´s best-selling smart phone, the iPhone.