成像光谱仪图像条带噪声去除的改进矩匹配方法  

Destriping Imaging Spectrometer Data by an Improved Moment Matching Method

作　　者：刘正军[1] 王长耀[1] 王成[1] 

机构地区：[1]中国科学院遥感应用研究所遥感信息科学重点实验室,北京100101

出　　处：《遥感学报》2002年第4期279-284,共6页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家 973项目 ;课题编号 :G2 0 0 0 0 7790 2

摘　　要：条带噪声是影响线阵CCD成像质量的一个重要因子。特别是对于CCD质量要求较高的高光谱成像仪 ,往往由于硬件质量造成了许多通道中条带噪声的出现。分析了条带噪声形成的主要原因 ,比较了几种常用条带噪声去除方法及其局限性。在此基础上指出 ,标准的矩匹配方法改变了图像在成像行或列方向的均值分布 ,造成了一定的灰度畸变。这种情况对于在地物非均匀分布状况下成像的小幅图像尤其明显。着重提出和讨论了利用均值补偿法、傅里叶变换法、相关系数法结合矩匹配方法来近似恢复由入射辐射强度产生的均值分布 ,从而达到保持图像质量并有效去除条带噪声的目的。并对条带噪声去除前后图像质量做了定性定量的比较、评价。Striping is an important factor that influences image quality acquired by linear array CCD blocks.This may be more crucial for spectrometers because of the imperfect calibration of the detector characteristics and the necessity of higher CCD quality,which results in the most common striping.This paper firstly discussed the main reason causing stripes.Then we compared some striping removal algorithms and their limitations.Based on this consideration,we point out that the standard moment matching method changes the mean value distrbution of image in line or column arrangement.This is especially the case for small size images.We present some approaches to simulate and calibrate the image gray value distribution.The purpose of these methods we embraced here is to recover the truth of the mean value distribution of ground radiance.As an emphasis,we mainly discussed the theories and processes of three kinds of mean value fitting approaches:the mean value compensation method,Fourier transformation method and correlation method. After discussing the methods,we used an 890×770 size test image acquired by an imaging spectrometer to experiment our theories.Through combining moment matching with these post-processing techniques,we successfully reduced stripes and improved the image quality.Finally,we visually and quantitatively assessed the quality of the resulted images.

关 键 词：去除 高光谱成像仪 条带噪声 矩匹配 均值补偿法 傅里叶变换 相关系数法 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]