Omnidirectional Human Behavior Recognition Method Based on Frequency-Modulated Continuous-Wave Radar

doi:10.1007/s12204-024-2580-z

Abstract

Abstract: Frequency-modulated continuous-wave radar enables the non-contact and privacy-preserving recognition of human behavior. However, the accuracy of behavior recognition is directly influenced by the spatial relationship between human posture and the radar. To address the issue of low accuracy in behavior recognition when the human body is not directly facing the radar, a method combining local outlier factor with Doppler information is proposed for the correction of multi-classifier recognition results. Initially, the information such as distance, velocity, and micro-Doppler spectrogram of the target is obtained using the fast Fourier transform and histogram of oriented gradients - support vector machine methods, followed by preliminary recognition. Subsequently, Platt scaling is employed to transform recognition results into confidence scores, and finally, the Doppler - local outlier factor method is utilized to calibrate the confidence scores, with the highest confidence classifier result considered as the recognition outcome. Experimental results demonstrate that this approach achieves an average recognition accuracy of 96.23% for comprehensive human behavior recognition in various orientations.

Key words: frequency-modulated continuous-wave radar, omnidirectional human behavior recognition, histogram of oriented gradients, support vector machine, micro-Doppler spectrogram, Doppler - local outlier factor

摘要： 调频连续波雷达可以在非接触且保护隐私的前提下实现人体行为特征识别，但人体姿态与雷达的空间位置关系直接影响识别精度。为了解决人体在非正面面对雷达时行为识别准确率低的问题, 提出了一种局部离群因子与多普勒信息相结合的方法,应用于多分类器识别结果的修正。首先通过快速傅里叶变换和梯度方向直方图-支持向量机方法得到目标的距离、速度、微多普勒图等信息并进行初步识别，然后利用普拉特缩放将识别结果转化为置信度，最后使用多普勒-局部离群因子方法对置信度进行校正，以置信度最高的分类器结果作为识别结果。实验结果表明，该方法对人体行为全方位的平均识别准确率可以达到96.23%。

关键词: 调频连续波雷达，全方位人体行为识别，梯度方向直方图，支持向量机，微多普勒图，多普勒-局部离群因子

CLC Number:

Sun Chang, Wang Shaohong, Lin Yanping. Omnidirectional Human Behavior Recognition Method Based on Frequency-Modulated Continuous-Wave Radar[J]. J Shanghai Jiaotong Univ Sci, 2025, 30(4): 637-645.

References

[1] ZHANG N, LI H, YU J, et al. Research progress of radar technology for non-contact vital sign detection [J]. Semiconductor Technology, 2023, 48(1): 1-9 (in Chinese). [2] LI Z, QUAN J, JIN H. Mine personnel behavior recognition algorithm based on 3D-Attention and multi-scale [J]. Foreign Electronic Measurement Technology, 2023, 42(07): 95-104 (in Chinese). [3] LIU T, LU Z, SUN Y, et al. Working activity recognition approach based on 3D deep convolutional neural network [J]. Computer Integrated Manufacturing Systems, 2019, 26(8): 2143-2156 (in Chinese). [4] GUO Y, MENG W, FAN Y, et al. Wearable sensor data based human behavior recognition: A method of data feature extraction [J]. Journal of Computer-Aided Design & Computer Graphics, 2021, 33(8): 1246-1253 (in Chinese). [5] FENG W G, LIU R, ZHU M. Fall detection for elderly person care in a vision-based home surveillance environment using a monocular camera [J]. Signal, Image and Video Processing, 2014, 8(6): 1129-1138. [6] ZENG X Z, BáRUSON H S L, SUNDVALL A. Walking step monitoring with a millimeter-wave radar in real-life environment for disease and fall prevention for the elderly [J]. Sensors, 2022, 22(24): 9901. [7] CHEN J X, WANG T, FANG Z Y, et al. Research on elderly users' intentions to accept wearable devices based on the improved UTAUT model [J]. Frontiers in Public Health, 2023, 10: 1035398. [8] KARA A. Protecting privacy in remote-patient monitoring [J]. Computer, 2001, 34(5): 24-27. [9] SHAH S A, TAHIR A, LE KERNEC J, et al. Data portability for activities of daily living and fall detection in different environments using radar micro-Doppler [J]. Neural Computing and Applications, 2022, 34(10): 7933-7953. [10] TIAN Z, YANG L, FU C, et al. Human action recognition based on multi-antenna FMCW radar [J]. Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition), 2020, 32(5): 779-787 (in Chinese). [11] FENG X, LI W, HE Z, et al. Human posture recognition based on multi-dimensional information feature fusion of frequency modulated continuous wave radar [J]. Journal of Electronics and Information Technology, 2022, 44(10): 3583-3591 (in Chinese). [12] SHRESTHA A, LI H B, LE KERNEC J, et al. Continuous human activity classification from FMCW radar with Bi-LSTM networks [J]. IEEE Sensors Journal, 2020, 20(22): 13607-13619. [13] XU Z, YIN H, LIN J. Two-persons activity recognition method for FMCW radar based on spatial information clustering [J]. Journal of Fuzhou University (Natural Science Edition), 2020, 48(4): 445-450 (in Chinese). [14] NALCI D, AKGUL Y S. Human action recognition with raw millimeter wave radar data [C]//2022 International Congress on Human-Computer Interaction, Optimization and Robotic Applications. Ankara. IEEE, 2022: 1-5. [15] CHEN X, YE N, XU K, et al. FMCW radar human action recognition system based on EfficientNet model [J]. Computer Technology and Development, 2002, 32(9): 134-141 (in Chinese). [16] LIU K, ZHONG P, XU D, et al. Tangent-based rectified linear unit [J]. Journal of Computer Integrated Manufacturing Systems, 2020, 26(1): 145-151 (in Chinese). [17] ZHOU M, WANG Y. Application of improved adaptive algorithm in Harmonic detection [J]. Manufacturing Automation, 2015, 37(23): 95-98 (in Chinese). [18] SUN Y, AI J, QU L. A hand gesture recognition method based on continuous wave radar [J]. Telecommunication Engineering, 2021, 61(7): 815-820 (in Chinese). [19] GUO X, HU G, YANG X. Vehicle detection method based on Radar and visual feature fusion[J]. Internet of Things Technologies, 2022, 12(2): 7-11 (in Chinese). [20] Xu W, Zhou P, Zhang F, et al. Pedestrian recognition algorithm based on information fusion of visual and millimeter wave radar [J]. Journal of Tongji University (Natural Science), 2017, 45(Sup.1): 37-42 (in Chinese).