Tracking 3D Shapes in Noisy Point Clouds With Random Hypersurface Models
State of the art depth-sensors such as range cameras (time-of-flight, structured light, stereo) or laser rangefinder obtain three-dimensional point cloud data of a given real-world scene. Recently, multi-sensor setups have received increasing attention. Point clouds have become highly relevant for many real world applications, such as surveillance, target tracking, 3D reconstruction, tele-presence, and free viewpoint television. Due to the large data amount of raw data, a depth sensor network gathers over time, the fusion of noisy point clouds has become necessary.