基于多传感器的人体运动模式识别研究
|
摘要
当前,随着人们日常生活水平的提高、科学技术的进步和医疗卫生条件的改善,人类的平均寿命在不断增长,这也使得老龄人口在总人口中所占的比重在不断提高,这一现象被称之为人口的老龄化。随着社会老龄化程度的不断加深,老年人的日常健康监护也越来越引起人们的重视。伴随着年龄的增长,老年人身体的各项机能开始衰退,活动能力、应变能力下降,在日常生活中极易发生各种危险。而当今社会独生子女数量较多,加之生活节奏快,工作压力大,子女无法经常陪伴在老人身边,致使许多老人在日常生活中长时间处于无人照看的状态,导致了空巢现象的发生。当危险发生时,空巢老人往往无法及时进行求救并获得医疗救助,从而导致较为严重的后果。
为了减少这种危险的发生,有必要对老年人的日常活动进行实时监测和识别,以便在危险发生时能够及时发现,避免情况的进一步恶化。基于无线体域网的实时运动监测方法是该领域目前研究的热点问题之一。在总结领域相关研究的基础上,结合组建家庭健康网络等课题的需求,本文研发了一种便携式人体运动信息采集装置,设计了不同人体运动模式的数据采集实验,并以实验数据为基础提出了一种基于支持向量机的多种类运动模式识别方法,实现了对常见人体运动模式的识别。
本文主要进行了以下工作：
1)设计了一种便携式人体运动信息采集装置,该装置的感知器件包括一个三轴加速度传感器和一个三轴数字陀螺仪,这两种传感器都是基于微机电系统技术的,可以对人体运动过程中的加速度和角速度两类主要运动信息进行实时测量。
2)综合分析了四元数法求解姿态角和加速度法求解姿态角的优势及不足,并在此基础上将两种方法进行有机整合,提出了一种基于加速度校准的人体姿态角求解方法,使用该方法可以较为准确的求出人体在任意时刻的姿态角。
3)提出了一种基于支持向量机的人体运动模式识别方法。该方法以人体的姿态特征和加速度特征作为识别特征,利用支持向量机模型对不同的人体运动模式进行识别。实验结果表明该方法具有良好的识别性能。
With the continuous improvement of the level of people's daily lives, the progress of science and the improvement of health conditions, the average human life expectancy is also increasing, which also makes the proportion of aging population in the general population improve, this phenomenon is called the aging of the population. With the deepening of the extent of an aging society, people have paid more and more attention to old people's day-to-day health care. Along with the growth of the age, the ability of old people begins to decline, all kinds of dangers can easily occur in everyday life. Now many families have only one child, because of the fast pace of life and the work pressure, children can not always accompany the elderly side, resulting in a lot of elderly people in their daily lives is unattended for a long period of time, leading to the occurrence of the phenomenon of empty nest. When danger occurs, empty nesters are often unable to carry out for help and get the medical assistance, resulting in more serious consequences.
In order to reduce this risk, it is necessary to real-time monitor and identify the daily activities of old people. It can be found in time when danger occurs, to avoid the situation worse. Real-time motion monitoring methods based on WBAN is one of the hot issues in the field. On the basis of the summary of research in the field, combined with the demand for the formation of topics such as Family Health Network, a portable human motion information acquisition device is designed and then a data acquisition experiment of different human motion mode is organized. On the basis of the data from the device an effective motion pattern recognition algorithm based on support vector machine is put forward.
In this paper, the following work was done:
1) A portable human body information acquisition device is designed, the device includes a three-axis accelerometer and a three-axis digital gyro, and the two sensors can measure the human body movement acceleration and angular velocity in real time.
2) Comprehensive analysis of the strengths and weaknesses of the quaternion method and acceleration method for solving attitude angle, and based on this, a method based on acceleration calibration is put forward, the attitude angle of the human body at any time can be more accurately obtained with this method.
3) A human motion pattern recognition algorithm based on support vector machine is put forward. Using the body's posture characteristics and acceleration characteristics as recognition features, using support vector machine to identify different motion patterns, verification test results show that the method has a good recognition performance.
为了减少这种危险的发生,有必要对老年人的日常活动进行实时监测和识别,以便在危险发生时能够及时发现,避免情况的进一步恶化。基于无线体域网的实时运动监测方法是该领域目前研究的热点问题之一。在总结领域相关研究的基础上,结合组建家庭健康网络等课题的需求,本文研发了一种便携式人体运动信息采集装置,设计了不同人体运动模式的数据采集实验,并以实验数据为基础提出了一种基于支持向量机的多种类运动模式识别方法,实现了对常见人体运动模式的识别。
本文主要进行了以下工作：
1)设计了一种便携式人体运动信息采集装置,该装置的感知器件包括一个三轴加速度传感器和一个三轴数字陀螺仪,这两种传感器都是基于微机电系统技术的,可以对人体运动过程中的加速度和角速度两类主要运动信息进行实时测量。
2)综合分析了四元数法求解姿态角和加速度法求解姿态角的优势及不足,并在此基础上将两种方法进行有机整合,提出了一种基于加速度校准的人体姿态角求解方法,使用该方法可以较为准确的求出人体在任意时刻的姿态角。
3)提出了一种基于支持向量机的人体运动模式识别方法。该方法以人体的姿态特征和加速度特征作为识别特征,利用支持向量机模型对不同的人体运动模式进行识别。实验结果表明该方法具有良好的识别性能。
With the continuous improvement of the level of people's daily lives, the progress of science and the improvement of health conditions, the average human life expectancy is also increasing, which also makes the proportion of aging population in the general population improve, this phenomenon is called the aging of the population. With the deepening of the extent of an aging society, people have paid more and more attention to old people's day-to-day health care. Along with the growth of the age, the ability of old people begins to decline, all kinds of dangers can easily occur in everyday life. Now many families have only one child, because of the fast pace of life and the work pressure, children can not always accompany the elderly side, resulting in a lot of elderly people in their daily lives is unattended for a long period of time, leading to the occurrence of the phenomenon of empty nest. When danger occurs, empty nesters are often unable to carry out for help and get the medical assistance, resulting in more serious consequences.
In order to reduce this risk, it is necessary to real-time monitor and identify the daily activities of old people. It can be found in time when danger occurs, to avoid the situation worse. Real-time motion monitoring methods based on WBAN is one of the hot issues in the field. On the basis of the summary of research in the field, combined with the demand for the formation of topics such as Family Health Network, a portable human motion information acquisition device is designed and then a data acquisition experiment of different human motion mode is organized. On the basis of the data from the device an effective motion pattern recognition algorithm based on support vector machine is put forward.
In this paper, the following work was done:
1) A portable human body information acquisition device is designed, the device includes a three-axis accelerometer and a three-axis digital gyro, and the two sensors can measure the human body movement acceleration and angular velocity in real time.
2) Comprehensive analysis of the strengths and weaknesses of the quaternion method and acceleration method for solving attitude angle, and based on this, a method based on acceleration calibration is put forward, the attitude angle of the human body at any time can be more accurately obtained with this method.
3) A human motion pattern recognition algorithm based on support vector machine is put forward. Using the body's posture characteristics and acceleration characteristics as recognition features, using support vector machine to identify different motion patterns, verification test results show that the method has a good recognition performance.
引文
[1]穆光宗,张团.我国人口老龄化的发展趋势及其战略应对[J].华中师范大学学报(人文社会科学版),2011,50(5)：29-36
[2]赵祥欣.基于三维加速度传感器的跌倒监测研究[D].杭州：浙江大学硕士学位论文,2008
[3]M. N. Nyan, F. E. H Tay, A. W. Y Tan, et. al. Distinguishing fall activities from normal activities by angular rate characteristics and high-speed camera characterization[J]. Medical Engineering& Physics,2006,28:842-849.
[4]邱显杰.基于视频的三维人体运动捕获方法研究[D].北京：中国科学院计算技术研究所博士学位论文,2006
[5]卢璇.室内运动人体检测与跟踪算法的研究与实现[D].成都：电子科技大学硕士学位论文,2010
[6]钱锋,梅雪,林锦国.基于矩形骨架的人体动作识别[J].微计算机信息,2012,28(4)：100-102
[7]崔莉,鞠海玲,苗勇等.无线传感器网络研究进展[J].计算机研究与发展,2005,42(1)：163-174
[8]李建中,高宏.无线传感器网络的研究进展[J].计算机研究与发展,2008,45(1)：1-15
[9]Akyildiz I F, Su W, Sankarasubramaniam Y, et al. Wireless sensor networks: A survey[J]. Computer Networks:The International Journal of Computer and Telecommunications Networking,2002,38(4):393-422
[10]任丰原,黄海宁,林闯.无线传感器网络[J].软件学报,2003,14(7)：1282-1291
[11]Natarajan A, Motani M, de Silva B, et al. Investigating network architectures for body sensor networks[C]. Proc of the 1st ACM SIGMOBILE International Workshop on Systems and Networking Support for Healthcare and Assisted Living Environments (HealthNet' 07). New York:ACM,2007:19-24
[12]Yun D, Kang J, Kim J E, et al. A body sensor network platform with two-level communications [C]. Proc of IEEE Int Syrup on Consumer Electronics. Washington:IEEE,2007:1-6
[13]Jang C S, Lee D G, Han J W. A proposal of security framework for wireless body area network[C]. Proc of the IEEE Int Conf on Security Technology. Washington:IEEE,2008:202-205
[14]Cai Y, Tan J. Secure group communication in body area networks[C]. Proc of IEEE Int Conf on Information Acquisition (ICIA). Washington:IEEE, 2008:555-559
[15]M. N. Nyan, Francis E. H. Tay, E. Murugasu. A wearable system for pre-impact fall detection[J]. Journal of Biomechanics,2008,41(16): 3475-3481
[16]Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, et al. PerFallD:A pervasive fall detection system using mobile phones[C]. Proc.8th Intl. Conf. Pervasive Computing and Communications Workshops (PERCOM Workshops). Washington:IEEE,2010:292-297
[17]Bourke A., O'Brien J., Lyons G. Evaluation of a threshold-based tri-axial accelerometer fall detection algorithm[J]. Gait& Posture,2007,26(2): 194-199
[18]Nyan M. N. et al. Application of motion analysis system in pre-impact fall detection[J]. Journal of Biomechanics,2008,41(10):2297-2304
[19]Chen J, Kwong K, Chang D, et al. Wearable sensors for reliable fall detection[C]. Engineering in Medicine and Biology Society. Washington: IEEE,2005:3551-3554.
[20]Gaetano A, Alessandro T, Nicola C, et al. Development of a novel algorithm for human fall detection using wearable sensors [C]. IEEE SENSOR 2008 Conference. Washington:IEEE,2008:1336-1339
[21]陈雷,杨杰,沈红斌等.基于加速度信号几何特征的动作识别[J].上海 交通大学学报,2008,42(2)：219-222
[22]Sa-kwang Song, Jaewon Jang, Soojun Park. A Phone for Human Activity Recognition Using Triaxial Acceleration Sensor[C]. Consumer Electronics. Washington:IEEE,2008:1-2
[23]Yu-Jin Hong, Ig-Jae Kim, Sang Chul Ahn, et al. Activity Recognition using Wearable Sensor for Elder Care[C]. Second International Conference on Future Generation Communication and Networking. Washington:IEEE, 2008:302-305
[24]ZHEN-YU HE, LIAN-WEN JIN. Activity recognition from acceleration data using AR model representation and SVM[C]. International Conference on Machine Learning and Cybernetics. Washington:IEEE,2008: 2245-2250
[25]王昌喜,杨先军,徐强等.基于三维加速度传感器的上肢动作识别[J].传感技术学报,2010,23(6)：816-819
[26]Lin Bao, Intille S S. Activity recognition from user-annotated acceleration data [C]. Proc of the 2nd International Conference on Pervasive Computing. Berlin:Springer Berlin Heidelberg,2004:1-17
[27]Kern N., Schiele B., Schmidt A.. Multi-sensor activity context detection for wearable computing[C]. In Proc. EUSAI, LNCS, Eindhoven, The Netherlands, November,2003. Berlin:Springer Berlin Heidelberg,2003: 220-232
[28]钱华明,夏全喜,阙兴涛等.基于Kalman滤波的MEMS陀螺仪滤波算法[J].哈尔滨工程大学学报,2010,31(9)：1217-1221
[29]付梦印,邓志红,张继伟Kalman滤波理论及其在惯性导航系统中的应用[M].北京：科学出版社,2003
[30]吉训生,王寿荣,许宜申等.自适应Kalman滤波在MEMS陀螺仪信号处理中的应用[J].传感器与微系统,2006,25(9)：79-81
[31]李广辉.基于小波方法的陀螺仪信号处理系统研究[D].天津：天津工业大学硕士学位论文,2006
[32]Donoho D L, Johnstone I M. Ideal spatial via wavelet shrinkage[J]. Biometrika,1994,81(3):425-455
[33]张维强,宋国乡.基于一种新的阂值函数的小波域信号去噪[J].西安电子科技大学学报(自然科学版),2004,31(2)：296-300
[34]张莲,秦华锋,余成波.基于小波闽值算法的研究[J].计算机工程与应用,2008,44(9)：172-173
[35]雷芳.地磁陀螺组合的姿态敏感技术[D].太原：中北大学硕士学位论文,2009
[36]张春慧.高精度捷联式惯性导航系统算法研究[D].哈尔滨：哈尔滨工程大学硕士学位论文,2005
[37]B. J. Munro, J. R. Steele, G. M. Bashford, et al. A kinematic and kinetic analysis of the sit-to-stand transfer using an ejector chair:Implications for elderly rheumatoid arthritic patients[J]. Journal of Biomechanics,1998, 31(3):263-271.
[38]文耀锋.一种实时的跌倒姿态检测和心率监护系统的研究[D].杭州：浙江大学博士学位论文,2008
[39]Saunders JB, Inman VT, Eberhart HD. The major determinants in normal and pathological gait[J]. Journal of Bone and Joint Surgery,1953,35(1): 543-558
[40]Morris RW. Accelerometry-a technique for the measurement of human body movements[J]. Journal of Biomechanics,1973,6(6):729-736
[41]Ermes Miikka, Parkka Juha, Mantyjarvi Jani, et al. Detection of daily activities and sports with wearable sensors in controlled and uncontrolled conditions[J]. IEEE Transactions on Informational Technology in Biomedicine,2008,12(1):20-26
[42]Godfrey A., Conway R., Meagher D., et al. Direct measurement of human movement by accelerometry[J]. Journal of Medical Engineering & Physics, 2008,30(10):1364-1386
[43]Li Ming, Rozgic Viktor, Thatte Gautam, et al. Multimodal physical activity recognition by fusing temporal and cepstral information[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering,2010, 18(4):369-380
[44]Preece S.J., Goulermas J.Y., Kenney L.P.J., et al. A comparison of feature extraction methods for the classification of dynamic activities from accelerometer data[J]. IEEE Transactions on Biomedical Engineering,2009, 56(3):871-879
[45]Wang Ning, Ambikairajah E., Lovell N.H., et al. Accelerometry based classification of walking patterns using time-frequency analysis [C]. Engineering in Medicine and Biology Society. Washington:IEEE,2007: 4899-4902
[2]赵祥欣.基于三维加速度传感器的跌倒监测研究[D].杭州：浙江大学硕士学位论文,2008
[3]M. N. Nyan, F. E. H Tay, A. W. Y Tan, et. al. Distinguishing fall activities from normal activities by angular rate characteristics and high-speed camera characterization[J]. Medical Engineering& Physics,2006,28:842-849.
[4]邱显杰.基于视频的三维人体运动捕获方法研究[D].北京：中国科学院计算技术研究所博士学位论文,2006
[5]卢璇.室内运动人体检测与跟踪算法的研究与实现[D].成都：电子科技大学硕士学位论文,2010
[6]钱锋,梅雪,林锦国.基于矩形骨架的人体动作识别[J].微计算机信息,2012,28(4)：100-102
[7]崔莉,鞠海玲,苗勇等.无线传感器网络研究进展[J].计算机研究与发展,2005,42(1)：163-174
[8]李建中,高宏.无线传感器网络的研究进展[J].计算机研究与发展,2008,45(1)：1-15
[9]Akyildiz I F, Su W, Sankarasubramaniam Y, et al. Wireless sensor networks: A survey[J]. Computer Networks:The International Journal of Computer and Telecommunications Networking,2002,38(4):393-422
[10]任丰原,黄海宁,林闯.无线传感器网络[J].软件学报,2003,14(7)：1282-1291
[11]Natarajan A, Motani M, de Silva B, et al. Investigating network architectures for body sensor networks[C]. Proc of the 1st ACM SIGMOBILE International Workshop on Systems and Networking Support for Healthcare and Assisted Living Environments (HealthNet' 07). New York:ACM,2007:19-24
[12]Yun D, Kang J, Kim J E, et al. A body sensor network platform with two-level communications [C]. Proc of IEEE Int Syrup on Consumer Electronics. Washington:IEEE,2007:1-6
[13]Jang C S, Lee D G, Han J W. A proposal of security framework for wireless body area network[C]. Proc of the IEEE Int Conf on Security Technology. Washington:IEEE,2008:202-205
[14]Cai Y, Tan J. Secure group communication in body area networks[C]. Proc of IEEE Int Conf on Information Acquisition (ICIA). Washington:IEEE, 2008:555-559
[15]M. N. Nyan, Francis E. H. Tay, E. Murugasu. A wearable system for pre-impact fall detection[J]. Journal of Biomechanics,2008,41(16): 3475-3481
[16]Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, et al. PerFallD:A pervasive fall detection system using mobile phones[C]. Proc.8th Intl. Conf. Pervasive Computing and Communications Workshops (PERCOM Workshops). Washington:IEEE,2010:292-297
[17]Bourke A., O'Brien J., Lyons G. Evaluation of a threshold-based tri-axial accelerometer fall detection algorithm[J]. Gait& Posture,2007,26(2): 194-199
[18]Nyan M. N. et al. Application of motion analysis system in pre-impact fall detection[J]. Journal of Biomechanics,2008,41(10):2297-2304
[19]Chen J, Kwong K, Chang D, et al. Wearable sensors for reliable fall detection[C]. Engineering in Medicine and Biology Society. Washington: IEEE,2005:3551-3554.
[20]Gaetano A, Alessandro T, Nicola C, et al. Development of a novel algorithm for human fall detection using wearable sensors [C]. IEEE SENSOR 2008 Conference. Washington:IEEE,2008:1336-1339
[21]陈雷,杨杰,沈红斌等.基于加速度信号几何特征的动作识别[J].上海 交通大学学报,2008,42(2)：219-222
[22]Sa-kwang Song, Jaewon Jang, Soojun Park. A Phone for Human Activity Recognition Using Triaxial Acceleration Sensor[C]. Consumer Electronics. Washington:IEEE,2008:1-2
[23]Yu-Jin Hong, Ig-Jae Kim, Sang Chul Ahn, et al. Activity Recognition using Wearable Sensor for Elder Care[C]. Second International Conference on Future Generation Communication and Networking. Washington:IEEE, 2008:302-305
[24]ZHEN-YU HE, LIAN-WEN JIN. Activity recognition from acceleration data using AR model representation and SVM[C]. International Conference on Machine Learning and Cybernetics. Washington:IEEE,2008: 2245-2250
[25]王昌喜,杨先军,徐强等.基于三维加速度传感器的上肢动作识别[J].传感技术学报,2010,23(6)：816-819
[26]Lin Bao, Intille S S. Activity recognition from user-annotated acceleration data [C]. Proc of the 2nd International Conference on Pervasive Computing. Berlin:Springer Berlin Heidelberg,2004:1-17
[27]Kern N., Schiele B., Schmidt A.. Multi-sensor activity context detection for wearable computing[C]. In Proc. EUSAI, LNCS, Eindhoven, The Netherlands, November,2003. Berlin:Springer Berlin Heidelberg,2003: 220-232
[28]钱华明,夏全喜,阙兴涛等.基于Kalman滤波的MEMS陀螺仪滤波算法[J].哈尔滨工程大学学报,2010,31(9)：1217-1221
[29]付梦印,邓志红,张继伟Kalman滤波理论及其在惯性导航系统中的应用[M].北京：科学出版社,2003
[30]吉训生,王寿荣,许宜申等.自适应Kalman滤波在MEMS陀螺仪信号处理中的应用[J].传感器与微系统,2006,25(9)：79-81
[31]李广辉.基于小波方法的陀螺仪信号处理系统研究[D].天津：天津工业大学硕士学位论文,2006
[32]Donoho D L, Johnstone I M. Ideal spatial via wavelet shrinkage[J]. Biometrika,1994,81(3):425-455
[33]张维强,宋国乡.基于一种新的阂值函数的小波域信号去噪[J].西安电子科技大学学报(自然科学版),2004,31(2)：296-300
[34]张莲,秦华锋,余成波.基于小波闽值算法的研究[J].计算机工程与应用,2008,44(9)：172-173
[35]雷芳.地磁陀螺组合的姿态敏感技术[D].太原：中北大学硕士学位论文,2009
[36]张春慧.高精度捷联式惯性导航系统算法研究[D].哈尔滨：哈尔滨工程大学硕士学位论文,2005
[37]B. J. Munro, J. R. Steele, G. M. Bashford, et al. A kinematic and kinetic analysis of the sit-to-stand transfer using an ejector chair:Implications for elderly rheumatoid arthritic patients[J]. Journal of Biomechanics,1998, 31(3):263-271.
[38]文耀锋.一种实时的跌倒姿态检测和心率监护系统的研究[D].杭州：浙江大学博士学位论文,2008
[39]Saunders JB, Inman VT, Eberhart HD. The major determinants in normal and pathological gait[J]. Journal of Bone and Joint Surgery,1953,35(1): 543-558
[40]Morris RW. Accelerometry-a technique for the measurement of human body movements[J]. Journal of Biomechanics,1973,6(6):729-736
[41]Ermes Miikka, Parkka Juha, Mantyjarvi Jani, et al. Detection of daily activities and sports with wearable sensors in controlled and uncontrolled conditions[J]. IEEE Transactions on Informational Technology in Biomedicine,2008,12(1):20-26
[42]Godfrey A., Conway R., Meagher D., et al. Direct measurement of human movement by accelerometry[J]. Journal of Medical Engineering & Physics, 2008,30(10):1364-1386
[43]Li Ming, Rozgic Viktor, Thatte Gautam, et al. Multimodal physical activity recognition by fusing temporal and cepstral information[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering,2010, 18(4):369-380
[44]Preece S.J., Goulermas J.Y., Kenney L.P.J., et al. A comparison of feature extraction methods for the classification of dynamic activities from accelerometer data[J]. IEEE Transactions on Biomedical Engineering,2009, 56(3):871-879
[45]Wang Ning, Ambikairajah E., Lovell N.H., et al. Accelerometry based classification of walking patterns using time-frequency analysis [C]. Engineering in Medicine and Biology Society. Washington:IEEE,2007: 4899-4902