张量扩散滤波在极化SAR图像降噪中的应用  

Application of Tensor-diffusion Based Filter in Despeckle of Polarimetric Synthetic Aperture Radar Imagery

作　　者：孙盛[1] 刘仁峰[2] 温雯[1] 刘竹松[1] 

机构地区：[1]广东工业大学计算机学院,广州510006 [2]华中科技大学电子信息与通信学院,武汉430074

出　　处：《科学技术与工程》2015年第18期190-194,共5页Science Technology and Engineering

基　　金：国家自然科学基金项目(61202269);顺德区产学研合作项目(2013CXY09)资助

摘　　要：为完成对全极化SAR图像相干斑噪声的有效抑制,引入了张量扩散模型。分析并引入了规范化的张量扩散模型,并对扩散张量的设计进行了分析。结合全极化SAR图像数据的格式,对扩散方程进行了改造,提出了适用于协方差矩阵数据格式的滤波方程,并给出了算法的实施步骤。对AIRSAR数据集中的样本图像Flevoland进行了定性的、定量的实验,实验结果表明,新的算法能够有效的提高等效视数并保持图像的空间分辨率和SAR数据的极化属性。For accomplishing the despeckle of Polarimetric Synthetic Aperture radar imagery, a tensor-diffusion model is introduced. After analysation, the normalized tensor-diffusion model is brought in and the designment of diffusion tensor has been completed. Based on the data format of Polarimetric Synthetic Aperture radar imagery, a remolded version for the diffusion equations has been done. The final filtering equation is put forward, which is fit for the covariance matrix of dataset. Qualitative and quantitative experiments are performed on the sample image Flevoland in AIRSAR dataset. The experimental results show that the new proposed method can effectively enhance the equivalent number of looks and preserve the spatial details and meanwhile polarimetric properties.

关 键 词：相干斑 张量扩散方程 规范化的张量扩散方程 扩散张量 等效视数 

分 类 号：TN911.73[电子电信—通信与信息系统]