摘要
有效地检测出异常活动能够更好地为老年人独立生活提供辅助。针对提高异常活动识别的准确率的问题,提出一种基于本体匹配的异常活动识别方法。采用本体对智能家居中的场景以及用户的行为进行建模,通过本体推理实现底层简单行为到高层复杂活动(ADLs)的识别,进一步将识别出的高层复杂活动与预定义的场景活动进行匹配,从而判断是否产生异常。该方法在本体推理的基础上增加了本体匹配的过程来实现异常活动的识别,从而使得识别结果更加准确。最后,通过异常活动识别原型系统(AARS)验证了该方法的可行性和有效性。实验结果表明,该方法对异常活动识别的平均准确率达到94.1%。
Detecting abnormal activity effectively can provide better assistance for the elderly to live independently. In order to improve accuracy of abnormal activities recognition,we proposed an ontology matching-based activity recognition method. It uses ontology to model the scene of smart home and users ' actions,and realises the recognition ranging from the underlying simple actions to high-level complex activities( ADLs) by ontology reasoning. Then by further matching the identified high-level complex activities with the predefined scenariobased activities,it can determine whether or not an abnormal activity occurs. The method adds the process of ontology matching on the basis of ontology reasoning to achieve the recognition of abnormal activities so that the recognition results become more accurate. Finally,through abnormal activity recognition system( AARS) we verified the feasibility and effectiveness. Experimental result showed that the average accuracy of the recognition on abnormal activities by the method reached up to 94. 1%.
出处
《计算机应用与软件》
CSCD
2015年第12期278-282,325,共6页
Computer Applications and Software
关键词
本体
本体推理
本体匹配
异常活动识别
智能家居
Ontology
Ontology reasoning
Ontology matching
Abnormal activity recognition
Smart Home