Recognition of Eyes-free Handwriting Input for Pen and Touch Interfaces

Continuous handwriting input using the finger on small devices such as PDAs and mobile phones is difficult if not impossible. Also, handwriting input has always required significant attention from the user during the process of writing. In our previous work, we proposed the concept of ‘eyes-free writing’, a novel solution that addresses both the problems. In this paper, we describe the challenges from the perspective of recognizing eyes-free handwriting input and explain the architecture of our recognition system in detail. Recognition is performed using wordlevel Hidden Markov Models (HMM), and empirical evaluation of the recognition accuracy using English and Tamil handwriting datasets has shown promising results.