Research on the Recognition Method of Bridge Target in SAR Image

An algorithm for SAR images based on knowledge of bridge over water is proposed. According to the positional relationship between bridge and river, it is pivotal for recognition to extract river correctly. First, pretreatment for SAR images is necessary, including noise reduction and image enhancement. Second, region-based threshold segmentation, region growing and mathematical morphology are used to extract water areas. Third, the water contours are extracted and traced. Fourth, potential bridge points are searched among water contours and fitted into lines which form two edge lines of potential bridge. Finally, feature extraction of potential bridges is implemented and certain rules are set to determine whether potential bridge is genuine target. The experiments show that the algorithm presented by this paper is effective.