模拟退火法在SAR单视图像斑点噪声抑制中的应用研究

Speckle Removal for SAR Single-look Images by Simulated Annealing

本文针对合成孔径雷达 (SAR)单视图像的特点 ,对R .G .White采用模拟退火估计雷达散射截面积(RCS)的算法进行了分析和改进 ;构造了新的邻域结构模型 ,在该模型上定义了一种新的目标函数 ,提出了相应的新算法 .理论分析和对仿真SAR图像、真实SAR单视图像的处理结果表明 ,新算法在不影响均匀区域降斑效果的同时可以更好地保持图像的结构特征 ,辐射失真小 。

Simulated annealing presents an additional way to suppress speckle in Synthetic Aperture Radar (SAR) images.This was exemplified by the simulated annealing algorithm for SAR Radar Cross Section (RCS) estimation developed by R.G.White.However,analysis shows White's algorithm can be refined when it is applied to a SAR single look image.Taking account of the speciality of the single look SAR image,this paper proposes another neighborhood model and a new objective function based on the model,and then suggests a novel simulated annealing algorithm.Theoretical analysis and experimental results using simulated and real SAR single look images show that,the novel algorithm improves structure preservation without degrading the speckle reduction in homogeneous areas.Moreover,it causes less radiometric distortion and shortens time cost with satisfying results.