• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Radar sensor technology XV >Human activity classification using Hilbert-Huang transform analysis of radar Doppler data
【24h】

Human activity classification using Hilbert-Huang transform analysis of radar Doppler data

机译：基于雷达多普勒数据的希尔伯特-黄变换分析的人类活动分类

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The automatic identification of human activities has become an area of interest in recent years. Identifying human activities is useful in various applications, such as through-barrier identification of intruders and non-contact monitoring of patients in hospitals. Numerous methods of human activity classification have been proposed in the past, including the use of Artificial Neural Networks (ANNs) and Support Vector Machines (SVMs). Most research in this area thus far has utilized the Short-Time Fourier Transform (STFT) as a method of obtaining the feature vectors necessary for classification. In this paper, we propose the use of the Empirical Mode Decomposition (EMD) algorithm as an alternative approach for obtaining feature vectors from human micro-Doppler signals and utilize an SVM for classification. Since the micro-Doppler signature is unique to a specific activity, the EMD outputs can be utilized as feature vectors. By utilizing the EMD algorithm in conjunction with an SVM, binary classification of human activities have shown to yield accurate results. Because SVMs were originally developed to solve the binary classification problem, additional steps must be taken in order to extend the problem to identify multiple classes. In this paper, two methods for multi-class classification will be demonstrated and compared. The first method is the one-against-all approach and the second is a decision tree based approach. In both cases, a high degree of accuracy is achieved
机译：近年来，人类活动的自动识别已成为人们关注的领域。识别人类活动在各种应用中很有用，例如入侵者的直通屏障识别和医院患者的非接触式监视。过去已经提出了许多人类活动分类方法，包括使用人工神经网络（ANN）和支持向量机（SVM）。迄今为止，该领域中的大多数研究已将短时傅立叶变换（STFT）用作获得分类所需的特征向量的方法。在本文中，我们提出使用经验模式分解（EMD）算法作为从人类微多普勒信号中获取特征向量的替代方法，并利用SVM进行分类。由于微多普勒签名对于特定活动是唯一的，因此EMD输出可以用作特征向量。通过将EMD算法与SVM结合使用，人类活动的二进制分类已显示出准确的结果。由于SVM最初是为解决二进制分类问题而开发的，因此必须采取其他步骤才能将问题扩展为识别多个类。在本文中，将演示和比较两种用于多类分类的方法。第一种方法是“反对所有”的方法，第二种方法是基于决策树的方法。在这两种情况下，都可以达到很高的精度

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号