基于水体散射特性的SAR图像水体检测

基于SAR图像中水体后向散射系数服从Gamma分布的特点,提出了一种基于水体散射特性的自动化水体提取算法。研究重点包括:第一,利用高斯分布拟合地物后向散射系数,并结合改进的Gamma模型进行分析,确定最优阈值位置,构建优化准则,自适应迭代求解目标阈值。第二,引入含水体的目标块选择准则,剔除非目标块,对目标块运用所建模型获取阈值,取各目标块平均值作为全景图最优阈值,该过程大大缩短了运行时间,增加了阈值准确率,对SAR图像中水体的实时检测具有深刻意义。实验数据选用Radarsat-2获取的淮南地区单极化数据,实验结果表明该算法可以实现大幅图像中水体的快速、精确和自动化提取。

SAR Image Water Extraction based on Scattering Characteristics

Due to the backscattering amplitude of water follows Gamma distribution in SAR images,an automatic water extraction method is proposed based on water’s scattering characterstics.The proposed method mainly includes the following two aspects:first,Gaussian model is used to fit the image backscattering amplitude,and combine with the Gamma model,the optimal threshold is set.Then a optimization criteria is established to realize the adaptive iterative solution procedure.Second,with the introduction of the block selection criteria,the non\|target block is removed,and the optimal threshold for each target block is obtained.Then the average of the target blocks is considered as the optimal threshold of the whole image.This method greatly reduces the running time and increases the accuracy rate.Radarsat\|2 data of Huainan area is used to verify the method .The experimental results show that the method enables the extraction process of water in SAR image rapidly,accurately and automatically.