Relative motion trajectory tracking-based Wi-Fi gesture recognition method

Abstract

The invention discloses a relative motion trajectory tracking-based Wi-Fi gesture recognition method. The method comprises the following steps of: firstly, detailedly analyzing an error source included in a phase of channel state information (CSI) obtained by a system, and completing error elimination by utilizing estimated values of a slope and an intercept; secondly, putting forward a new data fusion-based iterative interference suppression algorithm by utilizing the error eliminated channel state information so as to complete interference elimination of reflected signals and direct signalsbetween a transmitter and a receiver; thirdly, capturing signals reflected by a motion gesture by the system on the basis of interference suppression, constructing a virtual antenna array by utilizingcontinuous time signals received in a time domain, estimating wave incoming directions of the signals by utilizing a two-dimensional space smooth MUSIC algorithm, and reconstructing a motion trajectory of the gesture; and finally, completing gesture recognition by utilizing a Gaussian kernel-based support vector machine classifier according to the reconstructed motion trajectory. Experiments andanalysis results prove that the iterative interference suppression algorithm is capable of effectively suppressing interference signals, and a virtual array technology is put forward, so that overheads of reception-end antennas can be greatly decreased on the basis of improving the resolution and then the reconstruction of gesture motion trajectories is completed; and meanwhile, the method is notonly capable of completing single-hand gesture recognition, but also capable of correctly completing double-hand gesture recognition, and can better satisfy the requirement for gesture recognition inman-machine interaction.

Description

technical field [0001] The invention belongs to gesture recognition technology, and the system specifically relates to a gesture recognition method based on relative motion trajectory reconstruction under a Wi-Fi system. Background technique [0002] Gesture recognition, as a research branch in Human Computer Interaction (HCI), has become an important tool in the computer field. Compared with traditional interaction methods, gestures, as a clearer and more vivid body language, contain more rich information and help users express their ideas more intuitively. Research on gesture recognition plays an important role in promoting the development of HCI. [0003] Most of the research on traditional gesture recognition is based on vision (Vision Based) and sensor-based (SensorsBased). Vision-based gesture recognition systems have low recognition accuracy in weak light conditions. At the same time, because such systems require auxiliary equipment such as cameras, they are prone t...

AI Technical Summary

Problems solved by technology

This type of system uses gestures to identify the impact of received signal strength and phase, but the background noise in the signal is large and the interference signal is strong, which makes the received signal granularity and low resolution, and can only complete single-handed gesture recognition.

Images