Bridge detection in polarimetric SAR images based on water area tracing

Bridges interfere with automatic segmentation and linking of waterways crossed by bridges. This paper describes an automatic bridge detection method using polarimetric synthetic aperture radar (SAR) based on tracing of known water areas. The water areas are first accurately extracted using a level set segmentation algorithm from polarimetric SAR images. All the water branches related to the bridge are then linked by the scanning and the tracing of the water areas. The feature points on the water branch contours are extracted to identify the bridge end points. The bridge bodies are then developed from the end points. The algorithm was tested using polarimetric SAR images from RADARSAT-2 over Singapore and Lingshui, China. The results show that the detection rate of this algorithm is very high. Even small bridges in large scenes are correctly identified by the algorithm.