基于辐射不变点的高光谱遥感传感器交叉定标方法  

Cross-calibration Method of Hyperspectral Remote Sensing Sensor Based on Radiation Invariant Points

作　　者：李映潭 杨辽[2] 王杰[1] LI Yingtan;YANG Liao;WANG Jie(School of Geographical Sciences,China West Normal University,Nanchong,Sichuan 637002,China;Xinjiang Institute of Geography and Ecology,Chinese Academy of Sciences,Urumqi 830011,China)

机构地区：[1]西华师范大学地理科学学院,四川南充637002 [2]中国科学院新疆生态与地理研究所,乌鲁木齐830011

出　　处：《航天器工程》2024年第1期75-83,共9页Spacecraft Engineering

摘　　要：针对遥感影像数据处理中的辐射定标问题,关键在于精确获取辐射标定系数,以确保遥感影像像元亮度(DN)值与地表辐射值之间的准确对应。文章引入了地表矿物粉尘源调查卫星一等B级(EMIT L1B)数据,结合先前光谱成像仪器的经验和方法,提出了一套高光谱遥感传感器交叉定标的方法,基于多卫星数据和自动化程序完成了高光谱卫星定标系数的获取并提高了辐射不变点选取的准确性与效率。基于地表矿物粉尘源调查卫星(EMIT)与资源一号02D卫星数据采集交叉点,通过辐射不变点回归得到一套准确的定标系数。同时,引入了哨兵二号(Sentinel-2)数据选取辐射不变点,选取方法为通过自动化的程序对比两个光谱之间的相似度,该方法进一步提高了选点的高效性和可靠性。使用标准误差和相关系数评价标定后波段间的误差和相关性,其中最大标准误差低于0.008,145个波段相关系数大于0.90,证明了交叉定标方法能够获得高质量的反射率光谱曲线,对提高影像数据的准确性和可用性具有借鉴意义。In addressing the radiometric calibration issue in remote sensing image data processing,the key lies in accurately obtaining radiometric calibration coefficients to ensure a precise correlation between image digital number values and terrestrial radiative values.The article introduces EMIT L1B data and building upon previous experiences and methods from spectral imaging instruments,proposes a cross-calibration method for hyperspectral remote sensing sensors.This method,utilizing multi-satellite data and automated processes,acquires calibration coefficients for hyperspectral satellites and enhances the accuracy and efficiency of invariant point selection for radiometric calibration.Utilizing the intersection points of data collected by EMIT and the ZY-1-02D satellite,accurate calibration coefficients were derived through regression analysis at radiometrically invariant points.Simultaneously,Sentinel-2 data are introduced for selecting radiometric invariant points,employing an automated program to compare the similarity between two spectra.This approach further enhances the efficiency and reliability of point selection.The calibration results are evaluated using standard errors and correlation coefficients to assess errors and correlations between bands.The maximum standard error is found to be below 0.008,with 145 band correlation coefficients exceeding 0.90.This demonstrates that the cross-calibration method can yield high-quality reflectance spectral curves,underscoring its importance in improving the accuracy and usability of image data.

关 键 词：交叉定标 资源一号02D卫星 地表矿物粉尘源调查卫星 高光谱 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]