Knowledge acquisition from aerial images

Determining the position of a surface object from an aerial mobile tracking device is one of the major tasks in the control of airborne automatic-guided vehicles (AGVs). A knowledge acquisition algorithm that is able to classify roads in serial images is proposed. The information is acquired by encoding object features, which are then classified through a rule-based system in a database. To aid in the segmentation a generalized chain coding algorithm is employed. The approach presented represents the information extraction module of a vision-driven intelligent control system used to identify and follow objects of interest (roads) in an airborne video image.<
>