ASSET-2: visual tracking of moving vehicles

One of the major problems in the development of a capable autonomous land vehicle (ALV) is that of determining the presence, positions and trajectories of moving objects in the ALV´s vicinity. The research summarized in this paper goes some way to solving this problem, with a system called ASSET-2-A Scene Segmenter Establishing Tracking version 2. ASSET-2 is a self-contained real-time vision system which receives a sequence of images taken by a video camera, (mounted, for example, on a moving vehicle) and segments each image into separately moving objects using image motion. The motion is found by tracking two-dimensional features over time, and object boundaries, are temporally filtered to achieve accurate results. ASSET-2 is fast, accurate and reliable. It requires no camera calibration and no knowledge of the camera´s motion