Gesthaar: An accelerometer-based gesture recognition method and its application in NUI driven pervasive healthcare

Abstract

Natural user interfaces that detect gestures from acceleration data require fast and highly accurate signal matching algorithms to provide fluid interaction. In this paper, we contribute a novel system that uses Haar transform and Support Vector Machines for accelerometer-based gesture recognition. We benchmarked it using the uWave gesture library and found around 99% recognition accuracy after tenfold cross validation. Our method is better than those currently in the literature in terms of both accuracy and computational complexity. It also eliminates the necessity of personalized training and template adaptation. High accuracy and low computational complexity makes the proposed method suitable for gesture recognition using any kind of accelerometer equipped device.