基于UKF的毫米波雷达手势识别算法研究

Research on Millimeter Wave Radar Gesture Recognition Algorithm Based on UKF

本文提出了一种基于无迹卡尔曼滤波(UKF)的毫米波雷达干扰抑制的手势识别方法。首先根据原始雷达信号设置的采样点与线性调频信号数量,估计了目标的距离与多普勒参数。之后针对实际场景中类目标干扰较多的情况,设计了一套完整的基于UKF的场景类目标抑制方法,接着利用卷积神经网络(CNN)对不同手势距离⁃多普勒特征谱图进行提取和识别。实验结果表明,该抑制方法有效地解决了类目标干扰给手势识别带来的困扰,手势识别的平均准确率为98.74%,经过抑制干扰算法后准确率相较于干扰抑制之前提升了7.29%。

In this paper,we proposed a millimeter⁃wave radar gesture recognition method based on unscented Kal⁃man filter(UKF).Firstly,the distance and Doppler parameters of the target are estimated according to the sampling points set by the raw radar signals and the number of LFM signals.After that,a complete set of UKF⁃based scene target⁃like interference suppression method is designed for the situation that there are many target⁃like interference in the actu⁃al scene.Then,convolution neural network(CNN)is used to extract and identify the distance⁃Doppler feature spectrum of different gestures.The experimental results show that the suppression method effectively solves the problem of gesture recognition caused by target⁃like interference.The average accuracy of gesture recognition is 98.74%,and the accuracy is 7.29%higher than that before interference suppression.