Imputing Landsat7 ETM+ with SLC-off image using the similarity measurement between two clusters

This paper proposes a new method for imputing incomplete image from Landsat7 ETM+ SLC-off based on the comparison between two clusters of an image. In this method, we propose a two-step imputation in which the first step is a tentative imputing of a missing image using a neural network. The second step is imputing missing values by comparing similarity between two groups of data before imputing these missing values. We compared our imputation technique for the missing data problem of Landsat 7 ETM+ with the most well-known methods: Kriging algorithms, Local Linear Histogram Matching(LLHM), Linear regression algorithms. From the experimental results, our algorithms obtained the greatest accuracy among the various methods for imputing missing values in Landsat7 ETM+ with SLC-off.