一种基于双维度滤波和自适应定长化的FMCW雷达手势识别算法研究

Research on a Gesture Recognition Algorithm for FMCW Radar Based on Bidimensional Filtering and Adaptive Fixed Length

本文提出了一种基于调频连续波(Frequency Modulated Continuous Wave,FMCW)雷达回波信号的手势识别算法:首先,提出一种双维度滤波算法,在距离和速度维度对手势回波信号进行滤波,有效地降低了系统的静态噪声;其次,将数据经过动目标检测(Moving Target Indicator,MTI)算法滤除时间维度噪声;然后,提出了时间自适应定长化的方法,在减少势信息损失的前提下保证了每个手势样本帧数的一致性;最后,建立距离多普勒网络(Range Doppler Net,RD-Net)进行训练分类.该算法在谷歌开源的deep-soli数据集中取得了98.28%的准确率,比数据集提出者的算法的准确率提升了11.11%.该算法在实时推理实验中取得了90.8%的准确率,具有更好的泛化能力.

This paper proposes a gesture recognition algorithm based on frequency modulated continuous wave(FMCW)radar echo signals.Firstly,a two-dimensional filtering algorithm is proposed to filter the gesture echo signals in the distance and speed dimensions,which effectively reduces the static noise of the system.Secondly,the data is filtered by the moving target indicator(MTI)algorithm to filter out the noise in the time dimension.Then a time-adaptive fixed-length method is proposed,which ensures the consistency of the frame number of each gesture sample on the premise of reducing the loss of gesture information.Finally,a range Doppler net(RD-Net)is established for training and classification.The al⁃gorithm achieved 98.28%accuracy in Google's open source deepsoli data set,which is 11.11%higher than the algorithm proposed by the data set.The algorithm achieves 90.8%accuracy in real-time reasoning experiments and has better general⁃ization ability.