A headband for classifying human postures

A real-time method using only accelerometer data is developed for classifying basic human static postures, namely sitting, standing, and lying, as well as dynamic transitions between them. The algorithm uses discrete wavelet transform (DWT) in combination with a fuzzy logic inference system (FIS). Data from a single three-axis accelerometer integrated into a wearable headband is transmitted wirelessly, collected and analyzed in real time on a laptop computer, to extract two sets of features for posture classification. The received acceleration signals are decomposed using the DWT to extract the dynamic features; changes in the smoothness of the signal that reflect a transition between postures are detected at finer DWT scales. FIS then uses the previous posture transition and DWT-extracted features to determine the static postures.