CE-1干涉成像光谱仪数据信息提取技术研究
|
摘要
针对西安光机所研制的嫦娥一号(CE-1)干涉成像光谱仪(IIM)数据信息提取技术展开研究。IIM数据对于完成分析月球表面矿物类型含量与分布,绘制月球表面岩石分布图等科学任务具有重要意义。首先讨论了数据优化技术,其次研究了针对IIM数据的光度校正和反射率反演技术,最后分析月表高光谱图像的矿物质信息提取技术。
首先讨论了数据优化技术。针对IIM数据中的条带噪声,在传统线性插值法基础上,提出结合了三阶插值的改进线性插值方法,该方法对噪声点定位后,根据图像中噪声点上下行相应点的数值差别是否超过预设阈值,确定是采用线性插值还是三阶插值算法,修正噪声点像素值。对比试验采用均值和标准偏差以及图像质量因子(IQ因子)作为评价标准。结果表明:改进的线性插值方法在去除IIM影像条带噪声方面明显优于传统线性插值法和频域滤波法。
CE-1干涉成像光谱仪获取的干涉图数据通过光谱复原后得到的是目标的光谱辐亮度,其包含了目标反照率、光照条件、地形等因素的信息,而光谱反射率是分析月表物质类型的主要因素,故必须去除地形、光照等干扰。文中通过比较各种光度校正及反射率反演算法的优劣,采用布朗大学的经验公式对其进行光度校正,取得了良好的效果。并采用了平场域法、利用太阳辐射光谱两种方法对图像进行反射率反演,将辐亮度值反演为绝对反射率值。结果表明:两种方法反演后的绝对反射率曲线是基本一致的。
由于仪器分辨率、月壤混和特性等原因,IIM获得的光谱数据实际是多种物质类型的混合光谱。要提取物质类型的含量信息,必须进行混合像元分解。文中讨论了各种混合像元分解算法,选择最小距离分类法、光谱角分类法作为IIM的影像分析模型,将该方法用于CE-1的光谱反射率数据,得到的结果现有的理论是基本一致的。
Research on CE-1 Imaging Interferometer (IIM) spectral information extraction technology is implemented. The instrument is developed by Xi'an Institute of Optics and Precision Mechanics to analyze mineral contents and distribution of different types and draw rock distribution map on lunar surface. Data optimization technology is discussed firstly and the Photometric correction and reflection inversion is researched then. At last, the mineral information distraction technology on lunar surface is analyzed.
Data optimization technology is discussed. Aiming at the striping noise existing in IIM data, the Fourier transform algorithm and the linear interpolation algorithm are experimentally utilized for strip noise removal in IIM images. A modified linear interpolation algorithm is proposed, which combines low rank algorithm with high rank algorithm. After noise location, high rank interpolation or linear interpolation is used by judging whether the pixel difference between the upper and lower lines goes up to the threshold. Comparison of mean values and standard deviations as well as the IQ (Image Quality) factor shows that this modified method is evidently superior to the linear algorithm.
Hyperspectral images transformed from Interferogram obtained by IIM contain information on reflectance, light condition, terrain and so on. Reflectance is the main factor to analyze mineral type on lunar surface. Therefore, interference caused by terrain and light condition should be removed. By summarize the advantages and disadvantages of several kinds of method of Photometric correction and reflection inversion, empirical formula of Brown University is used on photometric correction. Both the Flat-Field method and the method which utilize the solar radiation spectrum are used on reflectance, inversion in order to transform radiant brightness image into reflectance, image. Results show that the absolute reflectance, curves obtained by both two methods are basically identical.
The spectrum data obtained by IIM is actually mixed spectrum of various materials due to instrumental resolution, mixed lunar soil, and so on. Therefore, pixel unmixing is essential to distraction information on mineral type. Different kinds of algorithms of pixel unmixing are discussed and compared. The minimum distance and spectrum angle mapper method are chosen as analyzing model for IIM data. Results show that the mineral distribution in the image is basically identical to the theory known to people in nowadays.
首先讨论了数据优化技术。针对IIM数据中的条带噪声,在传统线性插值法基础上,提出结合了三阶插值的改进线性插值方法,该方法对噪声点定位后,根据图像中噪声点上下行相应点的数值差别是否超过预设阈值,确定是采用线性插值还是三阶插值算法,修正噪声点像素值。对比试验采用均值和标准偏差以及图像质量因子(IQ因子)作为评价标准。结果表明:改进的线性插值方法在去除IIM影像条带噪声方面明显优于传统线性插值法和频域滤波法。
CE-1干涉成像光谱仪获取的干涉图数据通过光谱复原后得到的是目标的光谱辐亮度,其包含了目标反照率、光照条件、地形等因素的信息,而光谱反射率是分析月表物质类型的主要因素,故必须去除地形、光照等干扰。文中通过比较各种光度校正及反射率反演算法的优劣,采用布朗大学的经验公式对其进行光度校正,取得了良好的效果。并采用了平场域法、利用太阳辐射光谱两种方法对图像进行反射率反演,将辐亮度值反演为绝对反射率值。结果表明:两种方法反演后的绝对反射率曲线是基本一致的。
由于仪器分辨率、月壤混和特性等原因,IIM获得的光谱数据实际是多种物质类型的混合光谱。要提取物质类型的含量信息,必须进行混合像元分解。文中讨论了各种混合像元分解算法,选择最小距离分类法、光谱角分类法作为IIM的影像分析模型,将该方法用于CE-1的光谱反射率数据,得到的结果现有的理论是基本一致的。
Research on CE-1 Imaging Interferometer (IIM) spectral information extraction technology is implemented. The instrument is developed by Xi'an Institute of Optics and Precision Mechanics to analyze mineral contents and distribution of different types and draw rock distribution map on lunar surface. Data optimization technology is discussed firstly and the Photometric correction and reflection inversion is researched then. At last, the mineral information distraction technology on lunar surface is analyzed.
Data optimization technology is discussed. Aiming at the striping noise existing in IIM data, the Fourier transform algorithm and the linear interpolation algorithm are experimentally utilized for strip noise removal in IIM images. A modified linear interpolation algorithm is proposed, which combines low rank algorithm with high rank algorithm. After noise location, high rank interpolation or linear interpolation is used by judging whether the pixel difference between the upper and lower lines goes up to the threshold. Comparison of mean values and standard deviations as well as the IQ (Image Quality) factor shows that this modified method is evidently superior to the linear algorithm.
Hyperspectral images transformed from Interferogram obtained by IIM contain information on reflectance, light condition, terrain and so on. Reflectance is the main factor to analyze mineral type on lunar surface. Therefore, interference caused by terrain and light condition should be removed. By summarize the advantages and disadvantages of several kinds of method of Photometric correction and reflection inversion, empirical formula of Brown University is used on photometric correction. Both the Flat-Field method and the method which utilize the solar radiation spectrum are used on reflectance, inversion in order to transform radiant brightness image into reflectance, image. Results show that the absolute reflectance, curves obtained by both two methods are basically identical.
The spectrum data obtained by IIM is actually mixed spectrum of various materials due to instrumental resolution, mixed lunar soil, and so on. Therefore, pixel unmixing is essential to distraction information on mineral type. Different kinds of algorithms of pixel unmixing are discussed and compared. The minimum distance and spectrum angle mapper method are chosen as analyzing model for IIM data. Results show that the mineral distribution in the image is basically identical to the theory known to people in nowadays.
引文
[1]浦瑞良,宫鹏.高光谱遥感及其应用.北京.高等教育出版社.2000:22-30
[2]王晋年,张兵,刘建贵,等.以地物识别和分类为目标的高光谱数据挖掘.中国图象图形学报,1999,11(4):957-964
[3]Peg Shippert,Introduction to Hyperspectral Image Analysis.Research Systems,Inc.
[4]薛彬,CE-1干涉成像光谱仪信息处理及应用研究.博士论文.中国科学院西安光学精密机械研究所,2006
[5]李泳泉,刘建忠,欧阳自远,等.月球表面岩石分布类型的分布特征:基于Lunar Prospector(LP)伽玛射线谱仪谈探测数据的反演[J].岩石学报.2007,23(06):1169-1174
[6]李瑞玲,刘建忠,李春来.月球探测计划中影像数据的格式.地球物理学进展.2006,21(4):1155-1160
[7]侯波,基于小波变换消除遥感图像噪声.博士论文.中国科学院遥感应用研究所,2002
[8]陈正超.中国DMC小卫星在轨测试技术研究[D].中国科学院遥感应用研究所,博士论文,2005.
[9]高连如,张兵,张霞,等.基于局部标准差的遥感图像噪声评估方法研究[J].遥感学报.2007,11(2).
[10]吴军,张万昌.MODIS影像条带噪声去除的自相关插值法[J].遥感技术与应用,2006,36(7):253-257
[11]王志平,杨建峰,薛彬,等.利用高光谱数据进行地物识别分类研究[J].光子学报,2008,37(3):562-565
[12]罗欣,郭雷,杨诸胜.基于可逆整数变换的高光谱图像无损压缩[J].光子学报,2007,36(8):1457-1462
[13]WEGENER M J.Destriping multiple sensor imagery by improved histogram matching.Int.J.Remote sensing,1999,11(5):859-875
[14]GADALLAH F L,CSILLAG F.Destriping multisensor imagery with moment matching.Int.J.Remote Sensing,2000,21(12):2505-2511
[15]杨金红,顾松山,程明虎.插值法在去除MODIS遥感影像条带噪声中的应用[J].气象科学,2007,27(6):604-609
[16]孙颖,张志佳.基于条带频域滤波的自适应条带噪声去除算法[J].仪表技术与传感器,2006,2:57-59
[17]陈劲松,邵芸,朱博勤.中分辨率遥感图像条带噪声的去除[J].遥感学报,2004,8(3):227-233
[18]罗会信,曹艳力.一种高效的图像插值算法[J].计算机与现代化,2004,5:19-21
[19]刘正军,王长耀,王成.成像光谱仪图像条带噪声去除的改进矩匹配方法[J],遥感学报,2002,2(4):280-283
[20]GIOVANI CORSINI,MARCO DIANI,THOMA WALZE.Striping Removal in MOS-B Data[J].IEEE.on Geoscience & Remote Sensing,2000,38(3):1439-1446
[21]胥涛.“利用反射光谱和热辐射探测月球表面物质的理论和方法”,博士论文.中国科学院地球化学研究所,2004
[22]李雄耀,王志杰,程安云.月表有效太阳幅照度实时模型[J].地球物理学报.2008,51(1):25-30
[23]王晋年,郑兰芬,童庆熹,等.成像光谱图象光谱吸收鉴别模型与矿物填图研究[J].环境遥感.1996,11(1):20-30
[24]北京星图环宇科技公司编著.ENVI遥感影像处理专题与实践.2005.
[25]陈文霞,陈安升,蔡之华,等.基于高光谱吸收特征参数的分类研究[J].计算机工程与应用.2008,44(28):230-241
[26]Nirmal Keshava and John F.Mustard.Spectral Unmixing.In:Proceedings,IEEE Signal Processing Magazine,January 2002
[27]施介岚,徐百辉,史天元,等.以光谱混合分析法进行台湾地区Master 影像之研究.第二十三届测量学术及应用研讨会.2004.
[28]Charles Ichoku,Arnon Karnieli.A review of mixture modeling techniques for sub-pixel land cover estimation[J].Remote Sensing Reviews,Vol.13: 161-186.
[29]吕长春,王忠武,钱少猛,等.混合像元分解模型综述[J],遥感信息,2003:55-60
[30]Marsh,S,E,Switzer,P,Kowalik,W S and Lyon,R J P.Resolving the percentage of component terrains within single resolution elements[J].Photogrammetric Engineering and Remote Sensing,1980,(8):1079-1086
[31]Wang F.Fuzzy supervised classification of remote sensing images[C].IEEE Transactions on Geoscience and Remote Sensing,1990,(1):194-201
[32]惠巍巍.高光谱混合像元的分解及地物分类的研究,博士论文.东北林业大学.2007年6月
[33]许殿元,丁树柏.遥感图像信息处理.北京.宇航出版社.121-141
[2]王晋年,张兵,刘建贵,等.以地物识别和分类为目标的高光谱数据挖掘.中国图象图形学报,1999,11(4):957-964
[3]Peg Shippert,Introduction to Hyperspectral Image Analysis.Research Systems,Inc.
[4]薛彬,CE-1干涉成像光谱仪信息处理及应用研究.博士论文.中国科学院西安光学精密机械研究所,2006
[5]李泳泉,刘建忠,欧阳自远,等.月球表面岩石分布类型的分布特征:基于Lunar Prospector(LP)伽玛射线谱仪谈探测数据的反演[J].岩石学报.2007,23(06):1169-1174
[6]李瑞玲,刘建忠,李春来.月球探测计划中影像数据的格式.地球物理学进展.2006,21(4):1155-1160
[7]侯波,基于小波变换消除遥感图像噪声.博士论文.中国科学院遥感应用研究所,2002
[8]陈正超.中国DMC小卫星在轨测试技术研究[D].中国科学院遥感应用研究所,博士论文,2005.
[9]高连如,张兵,张霞,等.基于局部标准差的遥感图像噪声评估方法研究[J].遥感学报.2007,11(2).
[10]吴军,张万昌.MODIS影像条带噪声去除的自相关插值法[J].遥感技术与应用,2006,36(7):253-257
[11]王志平,杨建峰,薛彬,等.利用高光谱数据进行地物识别分类研究[J].光子学报,2008,37(3):562-565
[12]罗欣,郭雷,杨诸胜.基于可逆整数变换的高光谱图像无损压缩[J].光子学报,2007,36(8):1457-1462
[13]WEGENER M J.Destriping multiple sensor imagery by improved histogram matching.Int.J.Remote sensing,1999,11(5):859-875
[14]GADALLAH F L,CSILLAG F.Destriping multisensor imagery with moment matching.Int.J.Remote Sensing,2000,21(12):2505-2511
[15]杨金红,顾松山,程明虎.插值法在去除MODIS遥感影像条带噪声中的应用[J].气象科学,2007,27(6):604-609
[16]孙颖,张志佳.基于条带频域滤波的自适应条带噪声去除算法[J].仪表技术与传感器,2006,2:57-59
[17]陈劲松,邵芸,朱博勤.中分辨率遥感图像条带噪声的去除[J].遥感学报,2004,8(3):227-233
[18]罗会信,曹艳力.一种高效的图像插值算法[J].计算机与现代化,2004,5:19-21
[19]刘正军,王长耀,王成.成像光谱仪图像条带噪声去除的改进矩匹配方法[J],遥感学报,2002,2(4):280-283
[20]GIOVANI CORSINI,MARCO DIANI,THOMA WALZE.Striping Removal in MOS-B Data[J].IEEE.on Geoscience & Remote Sensing,2000,38(3):1439-1446
[21]胥涛.“利用反射光谱和热辐射探测月球表面物质的理论和方法”,博士论文.中国科学院地球化学研究所,2004
[22]李雄耀,王志杰,程安云.月表有效太阳幅照度实时模型[J].地球物理学报.2008,51(1):25-30
[23]王晋年,郑兰芬,童庆熹,等.成像光谱图象光谱吸收鉴别模型与矿物填图研究[J].环境遥感.1996,11(1):20-30
[24]北京星图环宇科技公司编著.ENVI遥感影像处理专题与实践.2005.
[25]陈文霞,陈安升,蔡之华,等.基于高光谱吸收特征参数的分类研究[J].计算机工程与应用.2008,44(28):230-241
[26]Nirmal Keshava and John F.Mustard.Spectral Unmixing.In:Proceedings,IEEE Signal Processing Magazine,January 2002
[27]施介岚,徐百辉,史天元,等.以光谱混合分析法进行台湾地区Master 影像之研究.第二十三届测量学术及应用研讨会.2004.
[28]Charles Ichoku,Arnon Karnieli.A review of mixture modeling techniques for sub-pixel land cover estimation[J].Remote Sensing Reviews,Vol.13: 161-186.
[29]吕长春,王忠武,钱少猛,等.混合像元分解模型综述[J],遥感信息,2003:55-60
[30]Marsh,S,E,Switzer,P,Kowalik,W S and Lyon,R J P.Resolving the percentage of component terrains within single resolution elements[J].Photogrammetric Engineering and Remote Sensing,1980,(8):1079-1086
[31]Wang F.Fuzzy supervised classification of remote sensing images[C].IEEE Transactions on Geoscience and Remote Sensing,1990,(1):194-201
[32]惠巍巍.高光谱混合像元的分解及地物分类的研究,博士论文.东北林业大学.2007年6月
[33]许殿元,丁树柏.遥感图像信息处理.北京.宇航出版社.121-141