uWave: Accelerometer-based personalized gesture recognition and its applications

摘要: The proliferation of accelerometers on consumer electronics has brought an opportunity for interaction based gestures or physical manipulation the devices. We present uWave, efficient recognition algorithm such using a single three-axis accelerometer. Unlike statistical methods, uWave requires training sample each gesture pattern and allows users to employ personalized manipulations. evaluate large library with over 4000 samples collected from eight elongated period time vocabulary patterns identified by Nokia research. It shows that achieves 98.6% accuracy, competitive methods require significantly more samples. Our evaluation data set is largest most extensive in published studies, best our knowledge. also applications gesture-based user authentication three-dimensional mobile interfaces created gestures.