论文部分内容阅读
基于乘积模型的杂波统计建模是进行高分辨极化合成孔径雷达(POLSAR)图像非均匀区域杂波统计特性分析的有效方法,其核心在于纹理分量随机分布的类型选择.广义伽马分布(GΓD)是一种普适性很强的分布类型,Weibull分布、伽马分布、逆伽马分布等都是GΓD分布的特例,因此基于GΓD分布的图像乘性杂波建模既满足杂波建模的高精度要求,同时成为各种雷达杂波辨识的有效工具.本文首先推导了服从GΓD分布的随机变量的高阶矩特征及其对数累积量(MLC),利用纹理分量服从GΓD分布情形构建了乘积相干斑模型,得到了适用于POLSAR图像处理的L分布杂波多视处理的协方差矩阵的概率分布函数,同时推导了其高阶矩特征及其对数累积量,提出了基于对数累积量的L分布参数估计新方法.针对样本数较少的情况下对数累积量参数估计失效问题提出了基于混合矩(MME)的参数估计方法来解决.然后给出了不同分布的高阶矩和对数累积量,通过二三阶对数累积量关系图辨析了常用分布与L分布的内在关系,得到了L分布是目前乘积模型中适用范围较为广泛的统计分布的结论.最后用仿真数据验证了理论推导的正确性,并将基于对数累积量的参数估计方法与已有方法进行了比较,结果证明新参数估计方法具有更高的估计精度和运算效率;另外,还用实测数据进行了统计模型检验,其结果验证了理论推导的正确性.极化SAR中多视图像L分布杂波的统计建模及其参数估计方法为极化SAR目标检测和识别等领域的新技术研究提供了新手段.
Statistical modeling of clutter based on product model is an effective method to analyze the statistical characteristics of clutter in non-uniform region of high-resolution polarimetric synthetic aperture radar (POLSAR) image, the core of which is the type selection of random distribution of texture components. The generalized gamma distribution GΓD) is a very universal distribution type. Weibull distribution, gamma distribution and inverse gamma distribution are all special cases of GΓD distribution. Therefore, the image multiplicative clutter modeling based on GΓD distribution not only satisfies the requirements of clutter Accuracy requirements of the model, and become an effective tool for a variety of radar clutter identification.This paper first derives high-order moments and their logarithmic cumulants (MLC) of random variables subject to the GΓD distribution, using the texture components obeying the GΓD distribution The product speckle model was constructed, and the probability distribution function of the covariance matrix for L distribution clutter multi-view processing applied to POLSAR image processing was obtained. Meanwhile, the higher order moments and their logarithmic cumulants were deduced. A new method for parameter estimation of cumulant L distribution is proposed.Aiming at the problem of failure of parameter estimation of logarithm cumulant with a small number of samples, a parameter estimation method based on hybrid moments (MME) Then the higher order moments and logarithmic cumulants with different distributions are given. The relationship between common distributions and L distributions is analyzed by the logarithmic cumulant relational graphs of second and third order, and the L distribution is obtained from the current applicable range The conclusion of the statistical distribution is more extensive.Finally, the correctness of the theoretical derivation is verified by the simulation data and the parameter estimation method based on the logarithm cumulant is compared with the existing methods. The results show that the new parameter estimation method has a higher estimation Accuracy and computational efficiency.In addition, the statistical model test is also carried out with the measured data, the results verify the correctness of the theoretical derivation.The statistical modeling of L-shaped clutter distribution and its parameter estimation method for polarimetric SAR are polarization SAR target detection and recognition areas such as new technology provides a new means of research.