Robust 3D Surveillance
Technical Report,01 May 2016,31 Jan 2017
University of Texas at Dallas Richardson United States
Pagination or Media Count:
Major Goals 1. Robustness of 3D data sensed by LiDAR Light Detection and Ranging cameras. LiDAR cameras have longer range for sensing and provide data for applications in multiple fields such as self-driving cars, geography mapping that can be used for surveillance as well. We will carry out anti-forensic and forensic studies on LiDAR data using the STIR funding. The preliminary results from this study will help us address the concern expressed by the reviewer for the earlier proposals focus on using only Microsofts Kinect data. 2. Real-time Performance of Multi-camera 3D Meshing We will also employ the STIR funding to get preliminary results on real-time performance of 3D reconstruction approaches. As mentioned earlier, we have been able to achieve real-time performance by working in the depth image domain, and mapping it back to the 3D. We will continue with our efforts and get additional preliminary results to show the feasibility for handling multiple camera data. Accomplishments 1. For RGB-D cameras, we first presented a real-time anti-forensic 3D object stream manipulation framework to capture and manipulate live RBG-D data streams to create realistic imagesvideos showing individuals performing activities they did not actually do. The framework uses computer vision and graphics methods to render photo-realistic animations of live mesh models captured using the camera. Next, we conducted a visual inspection of the manipulated RGB-D streams just like security personnel would do by users who are computer vision and graphics scientists. The study shows that it was significantly difficult to distinguish between the real or reconstructed rendering of such 3D video sequences, thus clearly showing the potential security risk involved. Finally, we investigated the efficacy of forensic approaches for detecting such manipulations.
- Computer Systems Management and Standards