Hi Kchien, I'm not really understanding your description of the different images that you wish to compare. It sounds like you have 4 different files, each of which has had the gaps filled using a different method. And each of these files contains only one band? Or do they contain all of the 6 VIS-NIR bands? If it is only one band, which band is it? I'm not sure what criteria you would use to determine which of these corrections is the best. I think you might need to ask, "best for what?" So, then that would help you set up a test for each image, to see how well it does at that defined task. Or, if you are just trying to minimize the striping, you might be able to display each image, and just see which looks best. You also mentioned classifying the images. What would the classes be, in this case? I am not aware of any literature about what is the best gap fill method. But I bet there have been studies about that. Have you done a literature search? Regards, Peg

Hi Peg, Thanks for your quick reply. Let me explain the case. I have a SLC-off file and a SLC-on file. I am only using band 3 and band 4, but i separated them before gap filling. (at the end i will merge band 3 +4 to analyze the NDVI of the different gap fill methods). Yes basically the main point is to reduce the striping in the images. So now i have 4 different gap filled images (4 x 3band gap fills and 4 x 4band gap fills, for 3 different regions -> total of 24 images), and somehow i would like to compare them (only of the same region and same band, so 6times a 4 image comparison), which method works best. comparing visualy in fact is possible, but i thought there might be a special method or something, so at the end i could display values/numbers in a chart or table. yes i have done literature search, but this section wasn't mentioned anywhere (they just compare subsets and how they look like comparing to each other). So an option could be using a scatter plot to compare the images..but then i don't know how i can describe them, and why a scatterplot should be "better" than an other :/... Thanks for your invested time Kind regards, Kchien ps.: analyzing NDVI will be questioned later =)