Modeling metadata of CCTV systems and Indoor Location Sensors for automatic filtering of relevant video content

Abstract

Location sensors and Closed-circuit Television (CCTV) cameras are widely used for the surveillance of people, objects and areas. These devices (sensors, CCTV cameras, etc.) generate a large amount of heterogeneous data, making their analysis and management difficult and very time-consuming. In the context of video-surveillance, automatic extraction of the relevant information among the mass of multi-sources information produced by these systems could significantly reduce the investigation time and facilitate their analysis. In this paper, we propose an approach that combines data from Indoor Location Sensors and metadata from CCTV cameras to automatically retrieve relevant video segments during research of evidence accident or crime events in indoor environments. The proposed method consists in i) the reconstruction of the mobile device trajectories from indoor location sensors and ii) the identification of the CCTV cameras intersecting the reconstructed trajectories. To ensure industrial transferability, indoor location sensors data were embedded in a generic model of CCTV cameras metadata that instantiates the standard ISO 22311 (relative to digital video-surveillance contents). We provide an experimental evaluation demonstrating the utility of our approach in a real-world case. Results show that our method helps the CCTV operators to effectively retrieve the relevant video and drastically reduce the time of analysis.