|
| 基于手机传感器的用户行为识别 |
| User Activity Recognition Method Based on Mobile Phone Sensors |
| 投稿时间：2016-10-09 |
| DOI：10.16018/j.cnki.cn32-1650/n.201703012 |
| 中文关键词: 用户行为识别 方向传感器 三轴加速度传感器 决策树 SVM分类器 |
| 英文关键词: user activity recognition direction sensor three-axis acceleration sensor dicision tree SVM classifier |
| 基金项目:重庆市教委科学技术研究资助项目(KJ1603207) |
|
| 摘要点击次数: 4612 |
| 全文下载次数: 3690 |
| 中文摘要: |
| 基于手机传感器实现用户行为识别在健康监控、时间管理和个人喜好分析、资讯筛选和推送等方面的重要作用，研究一种基于手机三轴加速度传感器、方向传感器获取用户数据，采用SVM多分类方法中的决策树分类方法，在决策树各节点训练SVM分类器，用于识别静止、步行、奔跑、上楼梯和下楼梯等5种日常行为，进而实现对用户行为的识别。通过对不同实验者的交叉对比实验，识别准确率平均为91.65%，证明了这一方法的有效性。 |
| 英文摘要: |
| User behavior recognition based on mobile phone sensors plays an important role in health monitoring, time management and personal preference analysis, information filtering and pushing. A method is developed to obtain user data based on three axis acceleration sensor and directional sensor. Using the decision tree classification method which belongs to the SVM multi-classification method, the SVM classifier is trained at each node of the decision tree to identify five kinds of daily behaviors such as stagnation, walking, running, stair climbing and down. And then the identification of user behavior is achieved. Through the cross contrast experiment of different experimenters, the average accuracy of recognition is 91. 65%, which proves the validity of this method. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |