Autonomous monitoring of human and vehicle activities in a wide area is very important for military and civilian applications. Using distributed sensors, one can generate 3D information and enable reliable monitoring of human and vehicle movements over a wide area in dynamic environments. Critical issues to be addressed for the design and development of a distributed sensor based monitoring system include: cross sensor self-calibration, long range target detection and tracking, target 3D feature extraction, and 3D information based human/vehicle tracking and verification. In this project, we will develop a PC system that uses a proprietary temporal/spatial variance analysis for high accuracy moving target detection at a distance, an efficient template matching technique for moving target tracking and cross platform/sensor object matching, a novel 3D motion trajectory analysis based cross sensor self-calibration, sensor fusion techniques for reliable human and vehicle detection and tracking, and techniques for 3D articulation-robust human tracking and verification. In Phase I, we will evaluate the usefulness of these algorithms for wide area personnel and vehicular monitoring and tracking, and develop a prototype design of the monitoring system. In Phase II, we will build a PC based prototype monitoring system and perform extensive real scenario tests. The primary military applications for the proposed work are for long range wide area monitoring to support early warning, force and sensitive area protection. Civilian applications include airport security monitoring, 24/7 facility protection, and bank and home security systems.
Keywords: Moving Object Detection, Human/Vehicle Detection, Sensor Fusion, Surveillance, Distributed Sensors, Automatic Self-Calibration, Human Activity I
