基于智能手机面向老年人的行为识别技术

发布时间：2019-03-25 20:06

【摘要】：如今智能手机由于配备了大量的传感器、计算和存储资源,使得基于各种传感器的应用出现在各种领域,如个人健康、环境监测、社交网络。目前,智能手机集成了丰富的传感器,如加速度计,陀螺仪,全球定位系统,麦克风,相机,亮度传感器,Wi-Fi和蓝牙接口,为人体活动识别提供一个合适的平台系统,也为研究人员提供了更为简单、方便的传感器数据获取方案。本文针对集成在智能手机中的GPS模块和三轴加速度传感器,提出了利用GPS速度数据结合人体在活动状态下产生的三轴加速度数据对老年人的室内和室外的行为进行实时识别以达到对老年人的行为监护的目的。本文的重点工作如下:(1)总结了基于加速度传感器的行为识别的工作流程及研究方法,对人体行为识别的各个功能模块分别作了详细的介绍。针对如今已经普及的智能手机提出一种基于智能手机的面向老年人的行为识别方法,该方法基于智能手机中的加速度传感器和GPS模块提出了一种新的行为数据模型,并采用该数据模型采集人体的行为数据。(2)对数据的分割进行研究,并为本系统选取了合适大小的滑动窗口对人体行为数据进行分割,随后列出本系统所使用的特征值。最后本章设计相应的对比实验证明当分类模型采用包含GPS速度特征值的特征向量进行训练和识别时相比未包含GPS特征值的特征向量的准确率要高,特别在人体乘车行为的识别准确率上有大幅度的提升。(3)在本文提出的基于智能手机加速度传感器和GPS模块的数据模型和行为识别方法的基础上,设计并实现了人体运动模式实时识别系统。系统实现了实时从移动终端接收人体行为数据并在后台处理、分类得出当前用户所处的位置是室内还是室外,并且把用户活动模式归类为静坐、站立、行走、跑步、骑车、乘车这6种状态。
[Abstract]:Today smart phones are equipped with a large number of sensors, computing and storage resources, making sensor-based applications in a variety of areas, such as personal health, environmental monitoring, social networks. Today, smart phones integrate a wealth of sensors such as accelerometers, gyroscopes, global positioning systems, microphones, cameras, luminance sensors, Wi-Fi and Bluetooth interfaces to provide an appropriate platform for human activity recognition. It also provides a simple and convenient sensor data acquisition scheme for researchers. In this paper, the GPS module and the three-axis acceleration sensor integrated into the smart phone, Using the GPS velocity data and the triaxial acceleration data produced by the human body in the active state, the real-time recognition of the behavior of the elderly in indoor and outdoor is proposed to achieve the purpose of monitoring the behavior of the elderly in order to achieve the purpose of monitoring the behavior of the elderly. The main work of this paper is as follows: (1) the workflow and research methods of behavior recognition based on acceleration sensor are summarized, and the functional modules of human behavior recognition are introduced in detail. Based on the acceleration sensor and GPS module in smart phones, a new behavior data model is proposed, which is based on the smart phone-oriented behavior recognition method for the elderly. The data model is used to collect the human behavior data. (2) the segmentation of the data is studied, and the sliding window of appropriate size is selected to segment the human behavior data, and then the characteristic values used in the system are listed. Finally, a comparative experiment is designed to prove that when the classification model is trained and recognized by the Eigenvectors containing the GPS velocity eigenvalues, the accuracy of the Eigenvectors without the GPS eigenvalues is higher than that of the classification models. Especially, the recognition accuracy of human ride behavior has been greatly improved. (3) on the basis of the data model and behavior recognition method based on smart phone acceleration sensor and GPS module proposed in this paper, A real-time human motion pattern recognition system is designed and implemented. The system realizes real-time receiving the human behavior data from the mobile terminal and processing it in the background, classifies whether the current user's position is indoor or outdoor, and classifies the user's activity mode as sit-in, stand, walk, run, and bike, and classify the user's activity mode as sitting, standing, walking, running, biking. Take a ride in these six states.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9

【参考文献】