Three-Dimensionalizing Surveillance Networks James Elder, Project Leader York University
The challenge To use persistent visual surveillance data to maintain and improve the security and efficiency of our urban centres in the face of rapid growth and increasing complexity
Current obstacles Most surveillance data are ignored due to lack of adequate manpower and reliable visual algorithms. Persistent visual surveillance systems are poorly integrated with other forms of geospatial information Surveillance cams compress the 3D scene into 2D, making inference difficult
Example (real, but anonymous public institution) 400 cameras 8 monitors 2 vigilant undergraduates What if something happens?
What if something happens?
A better way
Street level
Street Level
Goals Automatic, efficient, scaleable methods for extraction and integration of 2D and 3D urban data at street level Surveillance video, UAV photogrammetry, airborne & terrestrial LIDAR… Automatic inference of 3D scene properties Scene segmentation, building characteristics, foliage modeling Automatic inference of 3D scene dynamics Human and pedestrian traffic Integrated reporting and 3D visualization For efficient human interpretation Integration into distributed software architecture CAE S-Mission architecture
Scientific Questions There are many, e.g., How can 3D urban scene information be reliably extracted from single-view video? How can individuals be discriminated in crowds? How can free-form structures (e.g., trees) be reliably segmented from the scene? How can multiple forms of geolocation data (GPS, inertial, visual) be integrated to optimize positioning?
Applications Public and private security Urban planning Business analytics
Academic Team Claire Samson Carleton Frank Ferrie McGill Jim Little UBC Ayman Habib Calgary Dave Clausi John Zelek Waterloo York James Elder Gunho Sohn
Associated Korean Land Spatialization Group Projects Project 1. Real-time Aerial Monitoring System Project Leader: Impyeong Lee, Head, Dept. of Geoinformatics, The University of Seoul Project 2. Mobile Mapping at Street Level Project Leader: Taejung Kim, Associate Professor, Dept. of Geoinformatic Engineering, Inha University
Partners: 3D Modeling and Mapping
City of Toronto Survey & Mapping Services 2D and 3D mapping and modeling Asset management Bylaw enforcement
Defence Research & Development Canada 3D automatic target detection & recognition
CAE 3D modeling and simulation 3D immersive visualization Distributed real-time systems
Presagis COTS 3D modeling and simulation products
Applanix Mobile mapping and positioning GPS + Inertial + Visual
Array Systems 3D LIDAR scanning and modeling Scaleable signal processing systems
dmti Spatial Location-based data and services
Partners: Scene Dynamics
Ministry of Transport Ontario COMPASS Highway Surveillance Network
Honeywell Video Systems Intelligent visual systems for surveillance and business analytics People and object tracking Face detection Crowd density measurements
Aimetis Intelligent video surveillance systems Infrastructure Transportation Retail
Miovision Automated traffic flow analysis
Aeryon Labs Small electric UAVs for visual surveillance
For more on the project… Kick-off workshop Saturday 9-5 in King George Room: feel free to drop in.