Bright spot detection and classification method for a vehicular night-time video imaging system

摘要

Bright spots imaged by a forward-looking monochrome video camera during night-time operation of a host vehicle are detected and classified to determine the presence of leading and on-coming vehicles. A specified region-of-interest in each image frame is globally scanned to adaptively detect the bright spot contours, and search windows bounding the larger bright spot contours are locally scanned to adaptively detect individual bright spot contours that were fused in the global scan. A sensitive area within the region-of-interest is locally scanned to adaptively detect dim taillights of a leading vehicle, and path prediction of the host vehicle is used for frame-to-frame tracking of detected spots. Detected spots are classified depending on their location and frame-to-frame movement within the region-of-interest and their glare and pairing characteristics.