论文部分内容阅读
文中从参数估计角度研究基于l_k范数正则化的SAR复图像域快速自适应去噪方法。首先利用凸半二次正则化思想建立去噪模型与病态逆问题之间的联系,并依据迭代表达式进行参数估计方差与有偏CRB的比较,从而分析得到现有参数选择方法理论上的不足。然后对模型求解迭代表达式进行分析,得到包含正则化参数的模型解。继而应用最小化均方误差与单调有界数列原理,得到正则参数的选择方法与模型解的解析表达式,避免了求解的迭代过程,可以快速、自适应地实现去噪处理,并从理论上分析得到计算量减少的具体数值。最后研究了去噪模型对点目标分辨率的影响,建立了正则参数与分辨率的关系。仿真与实测SAR图像去噪结果验证了结论的正确性与方法的有效性。
In this paper, from the point of view of parameter estimation, fast adaptive noise reduction of SAR complex image based on l_k norm regularization is studied. Firstly, the relationship between the denoising model and ill-posed inverse problems is established by using the convex half-quadratic regularization and the comparison between the variance of the estimated parameters and the biased CRB according to the iterative expression, so as to find out the theoretical deficiencies of the existing parameter selection methods . Then, the iterative expression of model solving is analyzed and the model solution with regularization parameters is obtained. Then the principle of minimization of mean square error and monotonous and bounded series is applied to obtain the selection method of regularization parameter and the analytical expression of model solution, which avoids the iterative process of solution and can realize the denoising process quickly and adaptively. Analysis of the calculation to reduce the specific value. Finally, the effect of denoising model on the resolution of point target is studied, and the relationship between regular parameters and resolution is established. The results of simulation and actual SAR image denoising verify the validity of the conclusion and the validity of the method.