Translator Disclaimer
1 June 2015 Fusion of video and radar comparison to 3D ladar for activity recognition
Author Affiliations +
Determining hostile or suspicious activities within a civilian population can be challenging. Incorporating automated techniques for classifying activities can significantly reduce the operator workload. Utilizing 3D sensor modalities such as ladar can provide a strong capability for recognizing dismount activities. However, fusing multiple modalities, such as video in conjunction with radar, could provide a cheaper alternative for wide-area coverage. This work utilizes a single point-of-view 3D imaging system to approximate ladar captured data. Activity classification is done on the full 3D extracted motion, achieving 86% correct classification. Simulation of video-only activity classification is done by reducing the radial motion resolution and increasing the radial velocity error, and shows good performance on a significant number of activities. Simulation of radar-only classification is done by reducing the angular resolution and increasing the angular velocity error and shows good performance on a roughly orthogonal set of activities. Fusing the simulated radar and video data together at different fusion levels and comparing to the 3D ladar system gives an estimate of the loss in classification capability when using the less expensive fusion system.
© (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Dave Tahmoush "Fusion of video and radar comparison to 3D ladar for activity recognition", Proc. SPIE 9460, Airborne Intelligence, Surveillance, Reconnaissance (ISR) Systems and Applications XII, 94600B (1 June 2015);


Back to Top