Automatic gesture recognition is an important area of human–computer interaction. To date, recognition has mainly relied on video processing. This thesis uses smart watches, which contain motion sensors (built-in accelerometers) and offer wireless networking (mainly over Bluetooth), together with a computer or mobile phone running a motion recognition algorithm that turns the watch’s movements into predefined actions in the computer or phone interface. The algorithm will take as input the readings of the smart watch’s accelerometer and gyroscope and will use libraries to recognise the predefined movements, such as the $1 or $3 gesture recognizers (https://code.google.com/p/three-dollar-gesture-recognizer/).
