一种基于牛顿插值迭代的SAR影像滤波评价方法  

An Evaluation Method of SAR Image Filtering Based on Newton Interpolation Iteration

作　　者：张继超[1] 周沛希 刘鹏 ZHANG Ji-chao;ZHOU Pei-xi;LIU Peng(School of Geomatics,Liaoning Technical University,Fuxin 123000,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院

出　　处：《地理与地理信息科学》2019年第6期44-50,共7页Geography and Geo-Information Science

基　　金：国家863计划资助项目(2011AA120404)

摘　　要：针对SAR影像难以精确计算信噪比的问题,该文以滤波评价指标信噪比为依托,提出了一种基于牛顿插值迭代修正SAR影像近似信噪比的滤波质量评价方法:根据已知值通过牛顿插值得到更多的数据,再利用牛顿迭代法拟合非线性函数修正近似信噪比。以模拟影像和真实影像为数据源,选择目视效果较好的滤波结果作为真实影像的近似无噪声影像,人工对模拟影像和无噪影像进行降质,并使用5种滤波器对影像进行处理,分别利用基于牛顿插值迭代得到的修正近似信噪比、传统的峰值信噪比、等效视数和边缘保持指数评价滤波效果。结果表明:基于牛顿插值迭代修正的近似信噪比可作为衡量SAR影像滤波质量的评价指标,且无需参考原始无噪影像,对真实SAR影像的滤波质量评价有一定的可行性。Because of the hardness in retrieving an accurate signal to noise ratio(SNR)of SAR images,we propose a method to evaluate the filtering-quality of the approximate SNR adjusted by the process of Newton interpolation iteration.More data are generated by Newton interpolation from existing values,and then the approximate SNR is adjusted in terms of a non-linear fitting function constructed from the Newton iteration method.The simulated and actual images are used as the data sources,the filtering result with high visual confidence is chosen as the approximate noise-free image of the actual image,and the quality of both images is degraded artificially,then five filters are used to process the degraded images.Finally the filtering-quality is evaluated using the proposed method and the traditional PSNR approach,and the filtering effect is compared and analyzed in detail from qualitative and quantitative evaluation.The cross-comparison shows that results from both approaches are consistent,which demonstrates the potential of the proposed method in evaluating the filtering-quality of actual SAR images since the proposed method needn′t the noise-free baseline image.

关 键 词：SAR影像 牛顿插值迭代 滤波 信噪比 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]