SAR图像的绝对辐射精度直接影响SAR定量化应用水平，现有基于定标器的绝对辐射定标方法难以对SAR系统进行持续监测。如能在普通场景中找到一种稳定的散射特征量，则可将其作为参考实现常态化定标。针对C波段SAR图像，建立不同类别地物散射样本库，分析发现城区的后向散射系数中值重心具有良好的时间稳定性，进而提出一种基于神经网络的城区地物精筛与散射稳定特性提取方法。测试数据及与热带雨林数据的比较实验表明，该方法提取的散射稳定特性可用于SAR系统的常态化辐射定标。
The absolute radiometric accuracy of synthetic aperture radar (SAR) images directly affects the quantitative applications of SAR. It is difficult to continuously monitor SAR systems using conventional calibrator-based methods. However, if a stable backscattering feature can be observed in a common scene, its value can be used as a reference to ensure the routine calibration. In this study, a scattering sample database, which includes several different categories, is built using C-band SAR images, and the analysis of the results depicts that the median center of the backscattering coefficient in urban area is relatively stable over time. Further, a neural network-based method is presented and it can be used to finely filter the urban targets and extract stable backscattering feature. This method is validated by using the test data and a contrast experiment with rainforest data. This further illustrates that the stable feature extracted using this method can be used to perform routine radiometric calibration of SAR systems.
[1] Giudici D, Villa A, Recchia L, et al. Long term PS-CAL analysis of ERS and ASAR data and comparison to other calibration techniques[C]//Proceedings of the European Conference on Synthetic Aperture Radar. Berlin:VDE, 2014:1361-1364.
[2] 云日升, 郭伟. 亚马逊热带雨林星载SAR天线方向图获取与应用[J]. 测试技术学报, 2008, 22(4):301-306.
[3] Hawkins R, Attema E, Crapolicchio R, et al. Stability of Amazon backscatter at C-Band:spaceborne results from ERS-1/2 and RADARSAT-1[J]. Proceedings of the CEOS SAR Workshop, 2000, 450:99-108.
[4] Ridley J, Strawbridge F, Card R, et al. Radar backscatter characteristics of a desert surface[J]. Remote Sensing of Environment, 1996, 57(2):63-78.
[5] Horstmann J, Lehner S. A new method for radiometric calibration of spaceborne SAR and its global monitoring[C]//International Geoscience and Remote Sensing Symposium. New York:IEEE, 2002:620-622.
[6] 徐曦煜, 刘和光, 杨双宝. 星载雷达高度计沙漠散射特性及定标方法研究[J]. 遥感技术与应用, 2016, 31(5):893-899.
[7] 丁岩, 洪峻, 明峰, 等. 利用海面风场测量定标常数方法实验研究[J]. 国外电子测量技术, 2010, 29(6):68-71.
[8] Rizzoli P, Giudici D, D'Aria D, et al. Permanent scatterers for SAR sensor calibration[C]//European Conference on Synthetic Aperture Radar. VDE, 2008:1-4.
[9] D'Aria D, Ferretti A, Guarnieri A M, et al. SAR calibration aided by permanent scatterers[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010, 48(4):2076-2086.
[10] Iannini L, Guarnieri A M. A PS-based approach for the calilbration of spaceborne polarimetric SAR systems[C]//IEEE International Symposium on Geoscience and Remote Sensing. New York:IEEE, 2012:3297-3300.
[11] Schwerdt M, Schmidt K, Ramon N T, et al. Independent verification of the Sentinel-1A system calibration[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(3):994-1007.
[12] Schwerdt M, Schmidt K, Ramon N T, et al. Independent system calibration of Sentinel-1B[J]. Remote Sensing, 2017, 9(6):511-544.
[13] 李永晨, 刘浏. SAR图像统计模型综述[J]. 计算机工程与应用, 2013, 49(13):180-186.
[14] Tison C, Nicolas J M, Tupin F, et al. A new statistical model for Markovian classification of urban areas in high-resolution SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(10):2046-2057.
[15] 李松, 魏中浩, 张冰尘, 等. 深度卷积神经网络在迁移学习模式下的SAR目标识别[J]. 中国科学院大学学报, 2018, 35(1):75-83.
[16] 杜康宁, 邓云凯, 王宇,等. 基于多层神经网络的中分辨SAR图像时间序列建筑区域提取[J]. 雷达学报, 2016, 5(4):410-418.
