Extraction and recognition of bridges over water in high resolution SAR image

Abstract

Based on the characteristics of bridges over water in high resolution SAR image, a novel method for bridge extraction and recognition by combining the multi-scale decomposition and region analysis is proposed. Firstly, the nonsubsampled pyramid (NSP) transform is employed to denoise the SAR image. And then, by using the information provided by the multi-scale subbands and analyzing the region characteristics in the segmentation results, the contour of water region is extracted. Finally, the bridge is detected and recognized according to the knowledge of bridges over water. Experimental results obtained on real SAR images confirm the effectiveness of the proposed method. © Springer-Verlag Berlin Heidelberg 2011.