Multidimensional linear shift invariant velocity filters for vision-based automotive applications

Object detection and segmentation is one of the most challenging research topics in the field of active automotive safety systems. In order to warn the driver or automatically break before a potential collision, objects interfering the path of the host vehicle have to be detected and classified. Most recently developed approaches are based on two dimensional image processing, sometimes in combination with a tracking algorithm associating detections in consecutive frames to one and the same object. In contrast, the approach presented in this paper uses multidimensional velocity filters to identify moving objects, such as a pedestrian crossing the street in front of the vehicle. More precisely, movement characteristics (like velocity and direction) of objects in sight of a vehicle-mounted monochrome camera are used to enhance or suppress the corresponding pixels in the video stream. Basic considerations and example transfer functions for this application are presented and simulation results using a wave digital filter (WDF) realization thereof are presented and discussed.