Aiming at differences and instability of individual gestures signal

a continuous dynamic gesture recognition method based on acceleration sensor is proposed. The method utilizes MEMS acceleration sensor to capture gesture acceleration signal

according to the features of gesture signal

valid data of single gesture is automatically located and intercepted

after pre-processing and features extraction hidden Markov models（ HMM）to real-time identify specific gesture. A wearable gesture signal acquisition hardware prototype system is realized through design

1000 data of 10 types of gesture is selected to conducte the identification and comparative experiment

statistical results show that the method can identify continuous gesture in real-time and effectively.