| Synthetic Aperture Radar is one important means of microwaveremote sensing, playing a decisive role in marine surveillance. Shipdetection in SAR images can realize the traffic and maritime safetymonitoring and warning in fixed area because of its ability of all-weatherand all-time observation. And the efficiency of the monitoring is related tothe development of detection technology. So far, the false alarm rate of thetarget detection is so high that human intervention is still required for finaldecision.With the improvement of resolution, SAR images are closer to visualimage in human sight. To a great extent, rich details brought by highresolution provide great convenience to detection technology for furtherpromotion. Research of ship detection in high resolution SAR images hasbeen tremendous supported for further development.In this thesis, through the study of traditional ship detectiontechnology in SAR images, the framework of ship detection is summed upafter analysis. Then, the related algorithms and models are brieflyintroduced. And after the comparison of SAR image target characteristicsin different resolution, the similarities and differences are presented, andsome suggestions are put forward on the improvement for ship detectionbased on traditional methods. And also, key problems on ship detection inhigh resolution SAR images are summarized. After that, based on thetraditional CFAR-core method, introducing over-segmentation for spatialcorrelation extraction, Radiometric and Spatial Analysis based shipdetection in high resolution SAR images is presented. And theperformance of RSA based method is evaluated by comparing with traditional methods. Finally, a high performance software system of shipdetection in high resolution SAR image is introduced.The proposed method is proved to be effective and performed well inremoving false alarms after the experiments and analysis. And the softwarealso has good performance in the tests and does offer help to theresearchers. |
PDF Full Text Request