基于概率融合的人体行为识别方法
首发时间:2009-02-06
摘要:基于无线传感器网络的人体行为识别,特别是穿戴式行为识别,在环境智能、健康监护等众多领域应用非常广泛,受到了研究人员越来越多的重视。本文提出了一种基于概率融合的行为识别方法,通过对多个传感器数据分别建立识别模型,得到每个传感器对每种行为识别的概率,然后利用Dempster-Shafer证据理论对多个识别模型的概率估计进行融合,得到了比单个识别模型更准确、稳健的识别结果。基于真实行为数据的试验及分析证明了该方法的有效性和可行性。
For information in English, please click here
A Human Activity Recognition Method Based on Data Fusion with Probability Estimates
Abstract:The human activity recognition which based on wireless sensor networks has been researched and developed more and more, especially with the wearable devices. This paper introduce an human activity recognition method,based on data fusion with probability estimates.With this method, it classify the data from multi-sensors and get the probability estimates for each acitivities.Then,it fuse all probability estimates to one based on Dempster-Shafer evidence theory.The results of experiments indicate the efficiency of this method.
Keywords: Activity Recognition SVM D-S Evidence Theory Probability Estimates
基金:
论文图表:
引用
No.2850138449512338****
同行评议
共计0人参与
勘误表
基于概率融合的人体行为识别方法
评论
全部评论0/1000