Ship Detection With Polarimetric SAR Data

Ship detection has been widely used in the civil and the military. Polarimetric Synthetic Aperture Radar (PolSAR) can measure the polarimetric characters of objects and improve the accurate of ship detection and classification. The dissertation focuses on enhancing the contrast of ships to ocean and identifying the objects on the ocean surface based on the theory of PolSAR image and its engineering application background.The state-of-the-art of the study of ship detection with single and fully polarimetric SAR data is summarized. For dual-polarimetric SAR data, the two dimensional convolution is firstly proposed to detect ships. The algorithm improves the contrast of ships to ocean. For fully polarimetric SAR data, the improved generalized optimization of polarimetric contrast enhancement is presented to detect ships. The intensity of return wave is incorporated with the scattering mechanisms using the polarimetric feature vector. The above algorithms can detect ships when the intensity of ships approximates that of ocean. The fuzzy clustering is introduced in target identification on ocean surface. The man made targets, ships and ship wakes are identified with the unsupervised classification algorithm, which has great potential in targets identification on the ocean surface.The concept of polarimetric entropy is used in ship detection. Based on electromagnetic wave theory, the two kinds of entropy varying with incident angle, incident wave number and wind field are studied with Bragg and two-scale scattering model. It is shown that Cloude entropy is more adapt to ship detection than Touzi entropy with low wind.