Accurate Acoustic Based Gesture Classification with Zero Start-Up Cost
2018
Acoustic gesture recognition based on the Doppler effect has garnered much research attention. The accuracy of gesture recognition and potential false positives are the main factors that limit the widespread use of gestures. To this end, we propose a novel gesture classification method based on the acoustic Doppler effect that does not require any custom hardware, simply a speaker and one microphone on a laptop. An effective sound field is built by a high frequency sound wave from the speaker, and the wave reflected by hand motion is captured by the microphone. We design a set with five features, three of them are stable and invariant to different people, so even new users can operate our system with zero start-up cost and no training. The remaining two features are highly correlated with the velocity and the range to computer of the gestures, which can reduce the potential false positives in detection. Besides, a classifier is designed depending on multistage decision rules to identify the 11 kinds of defined gestures. The experiment result about user experience feedback of HCI shows that our system has good usability performance. And the numerical experiments with 10 users show that our system can not only keep less potential false positives, but also achieve a classification accuracy of up to 99.09%.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
21
References
2
Citations
NaN
KQI