Ontology And Dempster-Shafer Fused Human Activity Recognition

With the continuous development of pervasive computing and sensor technology,smart home system has been gradually close to people's lives.At present,user activity recognition technology has been applied to the field of smart home,and it is also one of the research hotspots nowadays.In the existing methods of human daily activity recognition,activities are uncertain due to many factors,such as missing sensor data,different execution order and different duration of activities.Ontology-based activity recognition is gaining interest due to its expressiveness and comprehensive reasoning mechanism,but its ability to deal with uncertain problems is poor,thus affecting the accuracy of activity recognition.To solve above-mentioned problems,an ontology-based and evidence-based human activity identification method is proposed in this paper.Although crisp ontology can promote knowledge sharing substantially,but it cannot deal with uncertainty.In this regard,evidence theory is a frequently-used basic uncertainty-reasoning fusion algorithm,especially used for multi-sensor data fusion.Accordingly,evidence theory is naturally employed to handle the uncertainty in activity recognition.Compared with the two independent methods,the fused method of their two can get more accurate results of activity recognition.Experiments show that ontology-and-evidence-theory fused method has high accuracy.The main work of this paper is as follows.1)In view of uncertainties in human daily activity recognition,a research framework based on ontology and Dempster-Shafer theory is proposed,Which can provide an overall process from human daily activity identification to uncertainty reasoning.2)In order to solve knowledge sharing problem caused by data heterogeneity in activity recognition process,the method of ontology-based activity modeling is adopted,which makes the activity ontology have a good reusability and be more easily to share.