基于独立成分分析的强噪声海域SAR图点目标提取-昆明理工大学学报.PDF

33 1 ( ) V ol 33 N o1 2008 2 Journal ofK unm ing U n iversity of Sc ience and T echnology ( Sc ience and T echnology) Feb 2008 SAR 1 2 2 刘芳宇 , 李昱彤, 杨 杰 ( 1 , 0 30024; 2 , 20024 0) : 将分形理论和独立成分分析( ICA, Indep enden t Com ponen tA na lysis)相结合, 用于强噪声海 域合成孔径雷达( SAR, Syn thetic A perture R adar) 图像的点目标提 首先依据分形理论, 计算点 态H lder指数, 再对指数图进行二值模糊处理, 接着将此图参与 ICA 计算, 然后使用泛化可调算 子收缩并优化其独立成分, 实现强噪声的有效抑制, 进而提 点目标 实验结果表明, H - ICA 与 基于 ICA 降噪的传统算法相比, 能够有效地降低海域强噪声, 并成功实现点目标提 : 目标识别; 合成孔径雷达; H lder指数; 雷达图象 : TP3914 : A : 1007 - 855X ( 2008) 01- 0023- 05 ICA - Based Point T arget Extract ion for Sea- SAR Im ages w ith Intens ive N oises 1 2 2 L IU F angyu , L I Y utong , YANG J ie ( 1 Co lleg e o f Science, T a iyu an U niversity of T echno logy, T a iyu an 030024, Ch ina; 2 In stitute o f Im age P ro cessing and P attern R ecognition, Shangh a i J iaotong U n iversity, Shangh a i 20024 0, Ch ina) Abs tract: F ractal and ICA ( Indep endent Com ponent A na lysis) are com b ined to detect po int targets in sea- SAR ( Synthet ic A p erture R adar) mi ages w ith inten sive no ises F irs,t po intw ise Hlder exponen t o f orig ina l mi age is computed Then, binary- fuzzy processing is em ployed to m odify the H lder mi age and the new H - mi age is u sed as input for ICA a lgorithm A generalized ad ju stab le op erator is u sed thereafter to enh ance the indep end ent components com puted from ICA, and the po int target can be detected in the recovery mi age fina l