Gesture recognition model based on 3D accelerations

This paper presents a recognition model on gesture interaction, gesture recognition based on accelerations, which are collected by 3D accelerometer. Unlike most other vision-based gesture recognition, our method resorting to accelerations of the gesture in three directions. Gesture is divided into small units, and each unit is modeled by an HMM, the HMM classifies the gesture unit after being well trained, the gesture identified when all of the corresponding gesture units recognized successfully, then the instruction will be triggered, and the human-computer interaction achieved. Through experiments, the method proved to be effective. This recognition model can be wildly used in the field of mobile computing and remote control, and people could use computer more friendly and naturally.