Abstract:

When the reflection interference of metal buildings is severe, a new coastline detection method is proposed based on edge detection mixing statistics and scattering characteristics. It overcomes the difficulty in accurate coastline detection by the traditional methods. The differences between statistical characteristics and scattering characteristics in adjacent region of the detection window are tested by using the likelihood ratio detector and ratio detector of volume scattering power respectively, and then the two detectors are fused by using the fuzzy set method. On the basis of coarse coastline detection by using the proposed detector, the coarse results are further improved by using the fast active contour model based on the edge energy map. The detection scheme is tested by using polarimetric synthetic aperture radar (SAR) images acquired by RADARSAT-2 over Dalian area of China. Experiment results demonstrate that the false edges resulted with the reflection interference of metal buildings and complex ocean environment are greatly reduced and the real coastline is accurately detected.