کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
535728 | 870369 | 2013 | 11 صفحه PDF | دانلود رایگان |
![عکس صفحه اول مقاله: Probabilistic human interaction understanding: Exploring relationship between human body motion and the environmental context Probabilistic human interaction understanding: Exploring relationship between human body motion and the environmental context](/preview/png/535728.png)
This paper presents an approach for modeling human interactions based on existent relationship characteristics between body parts motions and environmental parameters. Human interactions properly cannot be identified without knowing the relations between the objects such as human-robot and human–human. During any human interaction, there are many relations between human body parts and others. In this article a general model to analyse human interactions based on the existent relationships is presented. To study human motion properties, Laban Movement Analysis (LMA), a well-known human motion descriptor is used. This work focused onRelationship’s component of the LMA to analyse and formulate human activities related to environment. Bayesian approaches are proper classifiers for the mentioned goal, in order to be able to predict, define the existent dependencies, fuse different types of features and also deal with uncertainty. To present the idea, the model was performed to estimate some human movements and activities related to an object like a robot or another person. The result proves the capability of the approach to model and analyse any human activities related to environment using the LMA framework.
► Human interaction analysis by exploring in body motion-based relationship concept.
► Formulizing human relationship characteristics using a human movement descriptor.
► Relationship-based human movement analysis by a multilayer Bayesian Network (BN).
► Human activity analysis by probabilities of people’s movements related to a scene.
► The results showed the HMM approach are more reliable than BN in activity layer.
Journal: Pattern Recognition Letters - Volume 34, Issue 7, 1 May 2013, Pages 820–830