The application of multispectral analysis to reduce cloud interference

Abstract: For multispectral Remote Sensing (RS) image analysis, a big problem is that original dataalways include Clouds-Interference (CI). Especially in the bad weather conditions, the CI is evidentin RS image. So during the pre-processing of RS image, the CI should be reduced as much aspossible. In this paper, reducing CI is researched as the central problem, so that much Ground-Objects Feature (GOF) can be obtained. An analysis about the clouds reflection in differentSpectral-Bands (SBs) was done based on optical theory and early researches. Moreover, therelationships between clouds reflection and ground-objects reflection are presented to understandwhat the Digital Number (DN) represented in each SB, and to reduce the impact of CI the Same DNSpectral Matching Method (SDN-SMM) based on the multispectral application is applied. Finally,two cases are tested using Matlab Programme to indicate the rationality and practicability of SDNSMM.About SDN-SMM, some advantages and disadvantages are concluded through discussion onfinal results. The method can be used in any kind of multispectral sensors image with simplecalculation, while, the original data of clouds-free region will not be changed. However, the qualityof CI reduction depends on the precision of clouds identification and the SB which is used forspectral position relationship creating. In the end of this paper, the improvement is also presentedfor the future work.