Video scene analysis and irregular behavior detection for intelligent surveillance system
Sanghyuk park and Chang D. Yoo
This work considers a spatial-temporal hierarchical topic model for video scene analysis and irregular behavior detection in crowded traffic scenes for intelligent video surveillance. Previous probabilistic topic model algorithms, which are based on the bag-of-words representations of visual features, ignore the temporal dependencies of word occurrences. Thus, it is not suitable for analyzing sequential behaviors of the various objects in video sequences. The spatial-temporal hierarchical probabilistic latent semantic analysis (ST-HpLSA) is considered to describe spatial-temporal behavior patterns of moving objects in the crowded traffic scenes. The STHpLSA is able to detect the behaviors which occur both locally and globally over time and space. The ST-HpLSA is evaluated using the crowded traffic scene dataset. The experiments show that the ST-HpLSA yields good performance in analyzing behaviors and detecting irregular behaviors.
1. Sanghyuk Park and Chang D. Yoo, "Video scene analysis and irregular behavior detection for intelligent surveillance system", The 9th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI 2012), 2012.