Dual-Frequency Based Ultrasonic Gesture Recognition

In-air gesture recognition is becoming a popular means of interacting with terminal devices,such as smart phones,tablets and laptops.To implement in-air gesture recog-nition,many methods have been researched,such as computer vision based,RF based,WiFi based and ultrasonic based.Compared to other approaches,the ultrasonic based method shows superiorities in low-cost,robustness and so on.There are two ways to implement in-air gesture recognition,one of them directly analyses the features of sig-nal change casued by hand movement and the other tracks the trajectory of hand by positioning the hand.At present,most of studies utilize only single sound frequency,which limit the type of gestures,for example,they can not distinguish swipe left and swipe right ges-tures.In hand tracking,most algorithms have high computational complexity and rely on the specific hadrware.To solve these problems,in this dissertation,we construct two systems,one of them is the dual-frequency based ultrasonic gesture recognition system called DopGest and the other is the dual-frequency based two-dimensional hand trajec-tory tracking system.We mainly use the devices which already embedded in laptops and phones including two speakers and one microphone.Our systems work on the ul-trasonic environment.DopGest generates two different high frequency tones from two speakers and recognizes six gestures including single tap,double tap,swipe left,swipe right,swipe up and swipe down.For performance analysis,we test DopGest in out-doors and indoors,experimental results show that DopGest achieves about 86.3%aver-age gesture recognition accuracy.In two-dimensional hand trajectory tracking system,we continuously calculate the position of the hand relative to the speakers to achieve hand trajectory tracking.Our contributions are as follows:1)Most of ultrasonic based methods exploit single tone,which limit the type of gestures.For instance,they can not recognize horizontal swipe gestures.On the con-trary,DopGest can recognize swipe left and swipe right gestures by using two high tones(?18kHz).Also compared to computer vision based method,DopGest can work under any light conditions without extra hardware devices and with low power con-sumption.2)In IMU based solution,however,estimating distance traveled based on ac-celerometer requires double integration,and small measurement error can easily get exploded during double integration.Also,the measurement of accelerometer exists drift problem.Compared to IMU based method,Ultrasonic based method estimates velocity by calculating the Doppler frequency shift caused by movement without any specific hardware and reaches higher accuracy.