ALERT | Awareness and Localization of Explosive-Related Threats

Distributed Video Analytics and Anomaly Detection
F3-G

Download the 2012 Project Report

This project investigates the development of automated tools for video monitoring using multiple cameras in support of security applications. The project has both experimental and theo­retical components. In Year 4, we designed two novel methods to improve the performance of wide area video surveillance applications by using scene features. First, we evaluated the drift in intrinsic and extrinsic parameters for typical pan-tilt-zoom (PTZ) cameras, which stems from accumulated mechanical and random errors after many hours of operation. When the PTZ camera is out of calibra­tion, we show how the pose and internal parameters can be dynamically corrected by matching the scene features in the current image with a pre-computed feature library. Experimental results show that the proposed method can keep a PTZ camera calibrated, even over long surveillance sequences. Second, we introduced a classifier to identify scene feature points, which can be used to improve ro­bustness in tracking foreground objects and detect jitter in surveillance videos sequences. We show that the classifier produces improved performance on the problem of detecting counterflow in real surveillance video. We have also developed new approaches for anomaly detection based on charac­terization of anomalies using local spatio-temporal signatures that occur over a small window of time or space.