3D Intuitive Gesture Interaction via Motion Sensing

Motion sensing techniques are less limited in space and lighting from the point of view of human computer interaction.We employ accelerometers to study on how to effectively build natural and intuitive gesture-based user interfaces with machine learning methods.The gestures of the users are respectively recognized by hidden Markov model and support vector machine,in which the training data are semi-automatically generated by noise-adding techniques.It greatly reduces the workload of training and accordingly facilitates the development of gesture-based interaction.Experiments on a dataset of 70 gestures show that the average recognition rate is more than 90%.Two prototype interfaces,gesture-based phone-dialer and slide-presentation controller,are developed to verify the usability of our gesture-based interaction.