ABSTRACT: We developed a gesture based human computer interaction interface. Wireless sensor node is used to capture human body acceleration data. To segment received acceleration data stream, we developed an algorithm based on sliding window and standard deviation. To recognize gesture, Hidden Markov Model (HMM) which is a machine learning algorithm is used. Series prototype applications are built to demonstrate possible gesture based applications in future. We conducted several experiments as well. Finally, we got highest 96% accuracy and lowest 17% accuracy.
KEYWORDS: body sensor network, wireless sensor network, data stream processing, hidden markov model, machine learning, gesture recognition, human-computer interface
Source code will be available later