3D velocity filters for the detection of moving objects in image sequences

In the field of automotive applications, detecting an approaching vehicle or a crossing pedestrian, for example, is a complex problem that many research groups worldwide are dealing with. Most detection algorithms proposed in recent years are based on standard 2D image processing, sometimes combined with a tracking algorithm trying to associate detections in different frames to one and the same object. The approach presented in this paper is based on three-dimensional filters not only taking spatial but also temporal information into account. More precisely, moving objects in image sequences are enhanced or suppressed depending on their movement characteristics. Transfer functions for 3D filters working in a vision-based blind spot collision avoidance system are presented and discussed, and promising approaches for future application fields are proposed.