Human gesture recognition using bispectrum-based wireless signal processing
Research output: Contribution to journal › Article › Scientific › peer-review
|Number of pages||11|
|Journal||Telecommunications and Radio Engineering (English translation of Elektrosvyaz and Radiotekhnika)|
|Publication status||Published - 2020|
|Publication type||A1 Journal article-refereed|
Recently, state-of-the-art technologies are attracting considerable attention and developing successfully for human gesture detection, recognition and classification by using wireless signals. By using different hand motion in-air, a user could provide remote control of smart home devices or car multimedia systems. Systems operating without any physical contact between humans and different devices can be also useful for interface with computers exploiting of human gestures, gaming, outdoor lighting, remote controlling the UAV/multicopters, security applications, industrial robotics, health (vital sensing) and others. In these systems, detection, recognition and classification procedures are performed by extraction the contributions in the electromagnetic field caused by human gesture impact. In this paper, a novel bispectrum-based strategy is proposed and experimentally studied for human gesture classification. A novel type of classification features extracted from signal distorted by human gestures is suggested by evaluation of the third-order spectrum named bispectrum. It has been demonstrated that phase bispectrum or biphase contains unique discriminative features serving for the detection and classification of human gestures. The feasibility of developed hardware and software is demonstrated experimentally. It is shown that suggested bispectrum-based strategy provides invariance property to random signal time delays and considerable signal magnitude variations usually observed in the intricate indoor multi-path interference environment. Human gesture classification accuracy has been evaluated and discussed. Our results indicate the robustness of bispectrum-based information features for human gesture recognition and classification in a complicated indoor interference environment.