逆Gamma纹理背景下两类子空间目标的自适应检测方法

该文在复合高斯海杂波背景下,以逆Gamma分布作为纹理分量的先验分布模型,研究了1阶高斯（First Order Gaussian,FOG）和2阶高斯（Second Order Gaussian,SOG）两类子空间目标的自适应检测问题。采用两步广义似然比（Generalized Likelihood Ratio Test,GLRT）推导了检测统计量,并分别采用采样协方差矩阵（Sample Covariance Matrix,SCM）、归一化采样协方差矩阵（Normalized Sample Covariance Matrix,NSCM）和定点估计（Function Point Estimati...

Full description

Saved in:
Bibliographic Details
Published in雷达学报 Vol. 6; no. 3; pp. 275 - 284
Main Author 丁昊 王国庆 刘宁波 关键
Format Journal Article
LanguageChinese
Published 海军航空工程学院电子信息工程系 烟台 264001 2017
Subjects
复合高斯模型
子空间目标
自适应检测
逆Gamma纹理
自适应检测
逆Gamma纹理
子空间目标
复合高斯模型
Online AccessGet full text
ISSN2095-283X
DOI10.12000/JR16088

Cover

More Information
Summary:该文在复合高斯海杂波背景下,以逆Gamma分布作为纹理分量的先验分布模型,研究了1阶高斯（First Order Gaussian,FOG）和2阶高斯（Second Order Gaussian,SOG）两类子空间目标的自适应检测问题。采用两步广义似然比（Generalized Likelihood Ratio Test,GLRT）推导了检测统计量,并分别采用采样协方差矩阵（Sample Covariance Matrix,SCM）、归一化采样协方差矩阵（Normalized Sample Covariance Matrix,NSCM）和定点估计（Function Point Estimation,FPE）作为协方差矩阵估计值,与GLRT相结合,构造出新的自适应检测器。由于该文检测器在设计阶段考虑了海杂波的先验分布模型,且在检测阶段采用了与工作环境相匹配的模型参数,经性能分析与验证,其在检测性能上优于已有匹配滤波（Adaptive Matched Filter,AMF）和归一化匹配滤波（Adaptive Normalized Matched Filter,ANMF）检测器。
Bibliography:Adaptive detection; Compound Gaussian model; Inverse Gamma texture; Subspace targets
10-1030/TN
Ding Hao Wang Guoqing Liu Ningbo Guan Jian （Department of Electronic and Information Engineering, Naval Aeronautical and Astronautical University Yantai 264001, China）
Considering an inverse Gamma prior distribution model for texture, the adaptive detection problems for both first order Gaussian and second order Gaussian subspace targets are researched in a compound Gaussian sea clutter. Test statistics are derived on the basis of the two-step generalized likelihood ratio test. From these tests, new adaptive detectors are proposed by substituting the covariance matrix with estimation results from the Sample Covariance Matrix （SCM）, normalized SCM, and fixed point estimator. The proposed detectors consider the prior distribution model for sea clutter during the design stage, and they model parameters that match the working environment during the detection stage. Analysis and validation results indicate that the detecti
ISSN:2095-283X
DOI:10.12000/JR16088