Bridge Detection in Polarimetric SAR Images Based on Water Network Construction

A bridge detection method in polarimetric synthetic aperture radar (SAR) image based on water network construction using the graph model of Markov tree is proposed in this paper. First, branches of the water network are extracted by segmenting the sea and land using region-based level set segmentation method. All water branches are then connected by establishing the probabilistic graph model of Markov tree. Finally, the bridge bodies are recognized by merging the two contours of the adjacent branch pairs, which are represented by the edge of the Markov tree. Polarimetric SAR data acquired by RADARSAT-2 covered in Singapore is used for testing. Experimental results show that the method improves detection rate and false alarm compared to a traditional method.