Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)  2020, Vol. 54 Issue (2): 283-290    DOI: 10.3785/j.issn.1008-973X.2020.02.009
计算机技术、信息工程     
基于慢时间分割的超宽带雷达步态识别
周金海1(),王依川1,佟京鲆1,周世镒1,吴翔飞2
1. 浙江大学 信息与电子工程学院，浙江 杭州 310027
2. 杭州迈臻智能科技有限公司，浙江 杭州 310000
Ultra wide band radar gait recognition based on slow-time segmentation
Jin-hai ZHOU1(),Yi-chuan WANG1,Jing-ping TONG1,Shi-yi ZHOU1,Xiang-fei WU2
1. College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310027, China
2. Hangzhou Magnet Intelligent Technology Co. Ltd, Hangzhou 310000, China
 全文: PDF(1180 KB)   HTML
摘要：

为了解决室内监控时摄像头的隐私泄露问题和可穿戴设备的侵入性等问题，同时针对传统雷达步态识别算法中的行走条件限制问题，提出基于超宽带（UWB）雷达的自由空间步态识别算法.算法沿慢时间轴对目标行走动作产生的雷达步态信号进行分割产生一系列子信号；对于每个子信号，在距离单元上分别进行傅里叶变换得到距离-多普勒图像，这些距离-多普勒图像前后之间存在时序关系；利用方向梯度直方图算法对属于同一个步态信号的一组距离-多普勒图像进行特征提取，采用长短期记忆网络对得到的特征进行时序建模以获取目标身份分类结果. 实验在空旷的室内环境中进行，对四人的步态分类准确率为79.10%. 结果表明所提出的算法对自由空间中不同个体的步态具有一定的区分能力.
关键词： 超宽带UWB雷达;  自由空间;  慢时间;  分割;  方向梯度直方图;  长短期记忆网络    
Abstract:

An algorithm for free space gait recognition using ultra wide band (UWB) radar was proposed in order to solve the problems of privacy leakage caused by cameras and intrusiveness caused by wearable devices in the indoor monitoring, and to relax the restriction on walking conditions in traditional radar gait recognition algorithms. The radar gait signal reflected from the walking target is segmented along the slow-time axis to generate a series of sub-signals. For each sub-signal, the Fourier transform is performed on range bins to obtain a range-Doppler map. These range-Doppler maps temporally correlate each other. The features of a set of range-Doppler maps belonging to the same gait signal are extracted using the histogram of oriented gradient algorithm, and the resulting features are modeled by long short-term memory networks to obtain the final classification result of the target identity. The experiment was carried out in an empty indoor environment, and the accuracy of gait classification for four individuals was 79.10%. Results show that the proposed algorithm has a certain discrimination ability to different individual’s gait in free space.
Key words: ultra wide band (UWB) radar    free space    slow-time    segmentation    histogram of oriented gradient    long short-term memory network
收稿日期: 2019-07-20 出版日期: 2020-03-10
CLC:  TP 391.4  
基金资助: 国家自然科学基金资助项目（61675180，61575171）；浙江省基础公益研究计划资助项目（LGF20F020014）；浙江大学自主科研资助项目（H20151111）
作者简介: 周金海（1964—），男，实验师，从事微波光子学、智能传感技术、机器智能研究. orcid.org/0000-0003-1797-0399. E-mail： zhoujh@zju.edu.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章　　
周金海
王依川
佟京鲆
周世镒
吴翔飞

引用本文:

周金海,王依川,佟京鲆,周世镒,吴翔飞. 基于慢时间分割的超宽带雷达步态识别[J]. 浙江大学学报(工学版), 2020, 54(2): 283-290.

Jin-hai ZHOU,Yi-chuan WANG,Jing-ping TONG,Shi-yi ZHOU,Xiang-fei WU. Ultra wide band radar gait recognition based on slow-time segmentation. Journal of ZheJiang University (Engineering Science), 2020, 54(2): 283-290.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2020.02.009    或    http://www.zjujournals.com/eng/CN/Y2020/V54/I2/283

图 1  自由行走的时频图 (雷达频段：6.5~8.0 GHz)
图 2  步态信号的慢时间分割
图 3  典型步态信号的距离-多普勒图像
图 4  基于慢时间分割的步态识别算法框图
图 5  RDM的HOG特征提取
图 6  LSTM单元的信息流动关系
图 7  数据采集实验的环境配置
目标 身高/cm 体重/kg 目标 身高/cm 体重/kg
目标1 170 55 目标3 181 75
目标2 179 80 目标4 173 54
表 1  实验目标的物理特征
HOG LSTM
细胞尺寸 块尺寸 单元数 层数 准确率/%
8×8 3×3 32 1 76.45
2 75.52
64 1 77.26
2 77.66
128 1 79.67
2 78.86
12×12 2×2 32 1 70.70
2 70.23
64 1 71.64
2 71.04
128 1 71.71
2 71.30
表 2  HOG及LSTM模型参数对比
真实值 预测值 FNR
目标1 目标2 目标3 目标4
目标1 87.1% 2.9% 4.7% 5.3% 12.9%
目标2 5.1% 87.1% 5.2% 2.6% 12.9%
目标3 7.6% 6.9% 66.5% 19.0% 33.5%
目标4 14.0% 2.5% 8.3% 75.2% 24.8%
FPR 26.7% 12.3% 18.2% 26.9% ?
表 3  模型在测试集上的混淆矩阵
图 8  训练集和测试集的分类准确率和损失
1 WANG X Intelligent multi-camera video surveillance: a review[J]. Pattern Recognition Letters, 2013, 34 (1): 3- 19
doi: 10.1016/j.patrec.2012.07.005
2 YANG C C, HSU Y L A review of accelerometry-base-dwearable motion detectors for physical activity mon-itoring[J]. Sensors, 2010, 10 (8): 7772- 7788
doi: 10.3390/s100807772
3 AMIN M G, ZHANG Y D, AHMAD F, et al Radar signal processing for elderly fall detection: the future for in-home monitoring[J]. IEEE Signal Processing Magazine, 2016, 33 (2): 71- 80
doi: 10.1109/MSP.2015.2502784
4 AMIN M G. Radar for indoor monitoring: detection, classification, and assessment [M]. Boca Raton: CRC Press, 2017.
5 CHEN V C, LI F, HO S S, et al Micro-Doppler effect in radar: phenomenon, model, and simulation study[J]. IEEE Transactions on Aerospace and Electronic Systems, 2006, 42 (1): 2- 21
6 PARASHAR K N, OVENEKE M C, RYKUNOV M, et al. Micro-Doppler feature extraction using convolutional auto-encoders for low latency target classification [C]// 2017 IEEE Radar Conference. Seattle: IEEE, 2017: 1739-1744.
7 KIM Y, LING H Human activity classification based on micro-Doppler signatures using a support vector machine[J]. IEEE Transactions on Geoscience and Remote Sensing, 2009, 47 (5): 1328- 1337
doi: 10.1109/TGRS.2009.2012849
8 JOKANOVIC B, AMIN M Fall detection using deep learning in range-Doppler radars[J]. IEEE Transactions on Aerospace and Electronic Systems, 2017, 54 (1): 180- 189
9 SSYFIOGLU M S, OZBAYOGLU A M, GURBUZ S Z Deep convolutional autoencoder for radar-based classification of similar aided and unaided human activities[J]. IEEE Transactions on Aerospace and Electronic Systems, 2018, 54 (4): 1709- 1723
doi: 10.1109/TAES.2018.2799758
10 MURO A, GARCIA B, MENDEZ A Gait analysis methods: an overview of wearable and non-wearable systems, highlighting clinical applications[J]. Sensors, 2014, 14 (2): 3362- 3394
doi: 10.3390/s140203362
11 SAHO K, FUJIMOTO M, MASUGI M, et al Gait classification of young adults, elderly non-fallers, and elderly fallers using micro-doppler radar signals: simulation study[J]. IEEE Sensors Journal, 2017, 17 (8): 2320- 2321
doi: 10.1109/JSEN.2017.2678484
12 TIVIVE F H C, BOUZERDOUM A, AMIN M G A humangait classification method based on radar Doppler spectrograms[J]. EURASIP Journal on Advances in Signal Processing, 2010, 389716
13 SEIFERT A K, AMIN M, ZOUBIR A M Toward unobtrusive in-home gait analysis based on radar micro-doppler signatures[J]. IEEE Transactions on Biomedical Engineering, 2019, 66 (9): 2629- 2640
14 BARREET T W History of ultra wideband commun-ications and radar[J]. Microwave Journal, 2001, 44 (2): 22- 46
15 费元春. 超宽带雷达理论与技术[M]. 北京: 国防工业出版社, 2010.
16 张群, 胡健, 罗迎, 等 微动目标雷达特征提取、成像与识别研究进展[J]. 雷达学报, 2018, 7 (5): 5- 21
ZHANG Qun, HU Jian, LUO Ying, et al Research progresses in radar feature extraction, imaging, and recognition of target with micro-motions[J]. Journal of Radars, 2018, 7 (5): 5- 21
17 TSAO J, PORRAT D, TSE D Prediction and modeling for the time-evolving ultra-wide band channel[J]. IEEE Journal of Selected Topics in Signal Processing, 2007, 1 (3): 340- 356
doi: 10.1109/JSTSP.2007.906662
18 COHEN L. Time-frequency analysis: theory and applications [M]. Upper Saddle River: Prentice Hall, 1995.
19 ALAHI A, GOEL K, RAMANATHAN V, et al. Social lstm: human trajectory prediction in crowded spaces [C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 961-971.
20 VENUGOPALAN S, HENDRICKS L A, MOONEY R, et al. Improving lstm-based video description with linguistic knowledge mined from text [J/OL]. [2016-11-29]. (2019-06-25). https://arxiv.org/abs/1604.01729.
21 DALAL N, TRIGGS B. Histograms of oriented gradients for human detection [C]// International Conference on Computer Vision and Pattern Recognition. San Diego: IEEE Computer Society, 2005: 886-893.
22 SHIN H C, ROTH H R, GAO M, et al Deep conv-olutional neural networks for computeraided detection: CNN architectures, dataset characteristics and transfer learning[J]. IEEE Transactions on Medical Imaging, 2016, 35 (5): 1285- 1298
doi: 10.1109/TMI.2016.2528162
23 陈羽中, 方明月, 郭文忠, 等 基于小波变换与差分自回归移动平均模型的微博话题热度预测[J]. 模式识别与人工智能, 2015, 28 (7): 586- 594
CHEN Yu-zhong, FANG Ming-yue, GUO wen-zhong, et al Topic popularity prediction of microblog based on wavelet transformation and ARIMA[J]. Pattern Recognition and Artificial Intelligence, 2015, 28 (7): 586- 594
24 NARASIMHAN V, DANECEK P, SCALLY A, et al BCFtools/RoH: a hidden Markov model approach for detecting autozygosity from next-generation sequencing data[J]. Bioinformatics, 2016, 32 (11): 1749- 1751
doi: 10.1093/bioinformatics/btw044
25 CHEN K, ZHOU Y, DAI F. A LSTM-based method for stock returns prediction: a case study of China stock market [C]// 2015 IEEE International Conference on Big Data. Santa Clara: IEEE, 2015: 2823-2824.
26 NGUYEN A, KANOULAS D, MURATORE L, et al. Translating videos to commands for robotic manipulation with deep recurrent neural networks [C]// 2018 IEEE International Conference on Robotics and Automation (ICRA). Montreal: IEEE, 2018: 1-9.
[1] 王虹力,郭斌,刘思聪,刘佳琪,仵允港,於志文. 边端融合的终端情境自适应深度感知模型[J]. 浙江大学学报(工学版), 2021, 55(4): 626-638.
[2] 曾凌霄,刘宏伟,李伟,林勇刚,顾亚京. 基于变桨系统的功率与载荷双目标协同控制[J]. 浙江大学学报(工学版), 2021, 55(4): 750-756.
[3] 李文书,邹涛涛,王洪雁,黄海. 基于双尺度长短期记忆网络的交通事故量预测模型[J]. 浙江大学学报(工学版), 2020, 54(8): 1613-1619.
[4] 周登文,田金月,马路遥,孙秀秀. 基于多级特征并联的轻量级图像语义分割[J]. 浙江大学学报(工学版), 2020, 54(8): 1516-1524.
[5] 洪炎佳,孟铁豹,黎浩江,刘立志,李立,徐硕瑀,郭圣文. 多模态多维信息融合的鼻咽癌MR图像肿瘤深度分割方法[J]. 浙江大学学报(工学版), 2020, 54(3): 566-573.
[6] 张太恒,梅标,乔磊,杨浩杰,朱伟东. 纹理边界引导的复合材料圆孔检测方法[J]. 浙江大学学报(工学版), 2020, 54(12): 2294-2300.
[7] 梁栋,刘昕宇,潘家兴,孙涵,周文俊,金子俊一. 动态背景下基于自更新像素共现的前景分割[J]. 浙江大学学报(工学版), 2020, 54(12): 2405-2413.
[8] 赖小波,张学群,许茂盛. 胶质母细胞瘤多模态磁共振图像自动分割[J]. 浙江大学学报(工学版), 2019, 53(2): 355-363.
[9] 胡天中,余建波. 基于多尺度分解和深度学习的锂电池寿命预测[J]. 浙江大学学报(工学版), 2019, 53(10): 1852-1864.
[10] 郑洲, 张学昌, 郑四鸣, 施岳定. 基于区域增长与统一化水平集的CT肝脏图像分割[J]. 浙江大学学报(工学版), 2018, 52(12): 2382-2396.
[11] 穆晶晶, 赵昕玥, 何再兴, 张树有. 基于凹凸变换与圆周拟合的重叠气泡轮廓重构[J]. 浙江大学学报(工学版), 2017, 51(4): 714-721.
[12] 廖苗, 赵于前, 曾业战, 黄忠朝, 张丙奎, 邹北骥. 基于支持向量机和椭圆拟合的细胞图像自动分割[J]. 浙江大学学报(工学版), 2017, 51(4): 722-728.
[13] 付勇勇, 王旭航, 邓劲松, 叶自然, 周梦梦, 尤淑撑, 关涛. 采用国产GF-2遥感影像的复杂水网平原水体信息提取[J]. 浙江大学学报(工学版), 2017, 51(12): 2474-2480.
[14] 张建廷,张立民. 新型自适应稳健双边滤波图像分割[J]. 浙江大学学报(工学版), 2016, 50(9): 1703-1710.
[15] 吴一全,殷骏,朱丽,袁永明. 基于蜂群优化或分解的二维Arimoto灰度熵阈值分割[J]. 浙江大学学报(工学版), 2015, 49(9): 1625-1633.