中分辨率遥感图像条带噪声的去除  

Destriping CMODIS Based on FIR Method

作　　者：陈劲松[1] 邵芸[1] 朱博勤[1] 

机构地区：[1]中国科学院遥感应用研究所开放实验室,北京100101

出　　处：《遥感学报》2004年第3期227-233,共7页NATIONAL REMOTE SENSING BULLETIN

基　　金：中国科学院知识创新工程重要项目"数字地球基础理论研究"(KZCXZ 3 12 )

摘　　要：CMODIS是中国第一个发射上天的中分辨率成像光谱仪 ,它有 34个波段 ,波长范围从可见光到红外波段。CMODIS数据含有大量有很高实用价值的光谱信息。但是由于CMODIS传感器之间对接受的地物辐射信号的响应特性不同导致CMODIS数据中的许多波段含有大量的条带噪声 ,严重影响了CMODIS数据的解译和信息提取。在介绍了几种常用于TM ,MSS ,SPOT等多传感器遥感图像中的条带噪声去除方法基础上 ,提出了一种新的有限长脉冲响应滤波 (FIR)法 ,并比较了该方法和其它几种常用方法对几何纠正前后非均匀地物分布的CMODIS数据的去条带噪声结果。结果表明这种新方法要优于本文中提到的几种常用方法 ,具有很好的去条带噪声效果 ,同时保持图像原有信息。这种方法在其它多传感器遥感图像的条带噪声去除中也有很强的适用性。As the first Chinese moderate resolution spectrometer in SZ 3 spacecraft, CMODIS contain abundant spectral information with 34 channels in the range from visible to infrared. But sensor to sensor variation within instruments often leads to striping in many channels of CMODIS. The striping noise can distractingly and obstructively affect the interpretation and application of CMODIS data. This paper discusses the methods previously used in striping removal of TM,MSS ,MOS B and presents a new FIR method based on FFT transformation. The Application results of the method to geometrically corrected CMODIS data and non geometrically corrected CMODIS data and quantitative analysis of the results show that the new method can achieve a better result than the previously used methods mentioned in this paper in removing striping noise of CMODIS data and preserve the spectral characteristic of original image .The new method is also applicable in striping removal of other multisensor remote Sensing data.

关 键 词：CCD信号响应 矩匹配 低通滤波 傅里叶变换 条带去除 有限长脉冲响应滤波 

分 类 号：TP7511[自动化与计算机技术—检测技术与自动化装置]