Accelerometer-based gesture classification using principal component analysis

Gestures, such as a wave or a nod, are commonly used in daily lives. While gestures are most often used just as a support for our verbal communication, they can also be used as a sole, simple and effective way of communication. Recent developments in sensor technology, that have reduced the costs of small and precise sensors and allowed them to be built in a growing number of everyday devices, have also made it possible to explore and experiment with new modalities of communication in the area of human computer interaction. In the case of mobile devices, gesture-based interaction can be helpful for overcoming the physical size limitations, which make the usage of such devices particularly tedious. In this paper we propose a system that uses the accelerometer, embedded in a mobile phone, to capture simple gestures, such as hand describing a circle, thus allowing the user to draw or even write in the air. The principle component analysis is used for feature selection and dimensionality reduction in gesture classification. Experimental results are presented to demonstrate the efficiency of the proposed method.