Study On The Methods Of Adaptive Ship Detection In Spaceborne SAR Imagery

With the rapid development of spaceborne SAR(Synthetic Aperture Radar)and the increasing need of ship detection,research on adaptive ship detection in spaceborne SAR imagery is of very great importance.This dissertation focuses on practical problems of adaptive ship detection in Spaceborne SAR imagery,including analysis on the key factors influencing ship detection,adaptive ship detection and comprehensive ship discrimination in spaceborne SAR imagery.It occurs in a reliable and automated fashion.Analysis on the key factors influencing ship detection is the foundation of adaptive ship detection in spaceborne SAR imagery.The key factors are categorized into ocean factors,ship factors and SAR system factors.By analyzing these factors in deep,it illustrates in detail how ship detection is influenced and puts forward the most important factors influencing ship detection and discrimination.It lays the foundation for adaptive ship detection in spaceborne SAR imagery.The research on adaptive ship detection design and method is the core of adaptive ship detection in spaceborne SAR imagery.Based on the study mentioned above,this dissertation designs a scheme for adaptive ship detection in spaceborne SAR imagery,capable of processing a wide range of sensors,imaging modes and resolutions.Firstly,aiming at removing land in SAR imagery,this dissertation proposes an adaptive land masking method based on ship size and pixel size.Experimental results demonstrate that it can remove land efficiently.Secondly,by taking into account the imaging mode,swath width,polarization mode of SAR imagery and ship size,applying different strategies to SAR images in different resolution,it implements the adaptive ship detection in spaceborne SAR imagery.Experimental results based on RADARSAT-1,RADARSAT-2,TerraSAR-X,RS-1,RS-3 images validate that the scheme proposed in this dissertation is able to detect all potential ship targets in a fast,efficient and robust way.Finally,aiming at the drawback of the traditional CFAR(Constant False Alarm Rate)caused by only taking into account the intensity distribution of SAR imagery,this dissertation proposes a bilateral CFAR ship detection algorithm which combines the intensity and spatial distribution of SAR imagery.Experimental results demonstrate that it can reduce the influence of SAR ambiguities and sea clutter,acquiring better performance.Ship discrimination is the crux and difficulty of ship detection in spaceborne SAR imagery.Aiming at different types of typical false alarms in practical application,this dissertation proposes a ship comprehensive discrimination method in spaceborne SAR imagery based on complexity and confidence level.It categories discrimination based on features into certain discrimination and confidence discrimination based on features.Above all,targets which are false alarms certainly are identified by certain discrimination methods based on features.Then,we adopt confidence discrimination methods based on features to discriminate targets and figure out the confidence probability of each targets to represent the confidence level,followed by manual work to improve the reliability.Finally,this dissertation proposes to discriminate false alarms caused by azimuth ambiguities or ghosts based on RADARSAT-1 data.Taking into account different types of false alarms and combining automatic process and manual work,experimental results demonstrate that this method is able to discriminate false alarms efficiently.