【摘要】 基于恒虚警率的方法是经典的舰船检测算法,其核心是对海杂波统计分布的精确建模。随着合成孔径雷达(SAR)分辨率的提高,海杂波分布不再符合大数定理,而是表现为长拖尾形状,现已发展了诸如韦布尔分布、对数正态分布、G0分布、K-root分布、广义伽马分布等多种海杂波模型。本文首先利用上述模型对不同分辨率的SAR图像进行杂波建模,并在此基础上进行基于恒虚警率方法的舰船检测,然后,利用K-L距离和ROC曲线来评价各模型对不同分辨率条件下SAR海杂波拟合的精度与舰船检测的性能,给出了不同分辨率SAR适用于舰船检测的最优分布模型。更多还原

【Abstract】 The CFAR method is a classic ship detection algorithm,and its core is accurate modeling of sea clutter statistical distribution.With the increase of the SAR resolution,the sea clutter distribution no longer conforms to the large number theorem,it is presentthe long-tail shape The sea clutter based on Weibull distribution,lognormal distribution,G0 distribution,K-root distribution,and generalized gamma distribution models was developed.In this paper,the abovemodels are used to sea clutter modeling of SAR images at different resolutionsfirstly,and then a global CFAR detection is exploited.K-L distanceand ROC curves are used to evaluate the accuracy of models for sea clutter fitting and performance of ship detection and the optimal distribution model for ship detection with different resolution SAR is given.更多还原