MODIS影像条带噪声去除的自相关插值法

Destriping MODIS Images with Self-correlation Interpolation Algorithm

条带噪声是影响MOD IS影像质量和反演精度的一个重要因子。针对MOD IS第五波段影像条带特征,系统地分析了其形成的原因,比较了几种常用条带噪声去除方法并讨论了其具体应用的局限性。应用常用的傅立叶变换法进行了MOD IS条带去除试验,并根据MOD IS影像数据的自相关性提出了自相关插值法去除MOD IS影像条带噪声的方法。两种方法在MOD IS条带噪声去除实验结果的均值和标准偏差的比较表明自相关插值法在去除MOD IS影像条带噪声方面要明显优于傅立叶变换法。

Strip noise is one of the important factors affecting the quality of the imaginary acquired from space and the further processing on information retrieving from MODIS 1B image. Aiming at the characteristics of strip noise in Band 5 of MODIS 1B image, the cause of strip noise formation was described, and the commonly used methodologies and principles for destriping imaginaries from several different sensors were systematically compared with discussions focusing mainly on their constraints in operations. In this study, the Fourier Transform Algorithm, the most frequently used methodology for destriping, was experimentally utilized for strip noise removal in MODIS IB image, and in the light of selfcorrelation characteristic of MODIS data, a new approach by means of self-correlation interpolation algorithm for effective removal of strip noise in Band 5 of MODIS image was proposed. Comparison of mean values and standard deviations as well as edge affection obtained from the strip noise removed image by these two methodologies suggested that self-correlation interpolation algorithm is evidently superior to the traditional Fourier transform algorithm in destriping MODIS 1B products.