Feature-level fusion for free-form object tracking using laserscanner and video

A scalable feature-level sensor fusion architecture combining the data of a multi-layer laserscanner and a monocular video has been developed. The approach aims at a maximization of synergetic effects by combining low-level measurement features and at the same time trying to keep the fusion architecture as general as possible. A new concept for the geometric modeling of diverse object shapes found in real traffic scenes, including free form models, enhances the precision of the object tracking. Results from real sensor data demonstrate the performance of the new algorithms compared to robust algorithms known from the literature.