An Ontology for video human movement representation based on Benesh notation

Segmentation and representation of movements in videos play an important role in different applications such as search engines, video recommender systems, and video summarizers. In this paper, we present a system for semantic annotation of movements in video. This system is based on the temporal segmentation method that extracts the movement objects in still scenes, and on the high-level movement concepts to bridge the semantic gap between such concepts and the low-level video features. We propose a knowledge-based Model of movements in videos by using the OWL ontology and SWRL rules. Our Video Movement Ontology (VMO) considers different concepts related to the relevant movement features, which is based on the semantic of the Benesh Movement Notation (BMN). BMN can describe any form of dance or human movement. Rules in description logic are defined to describe how low-level features and mapping process between those features and ontology's concepts should be applied according to different perception of video content analysis. This system can improve the quality of annotation of movements in the videos and can discover the hidden information by reasoning video knowledge and movement's features.