Similar patch search in RS images using different metrics

An often used operation in many remote sensing (RS) applications is searching for similar patches in a given image or a set of images. For example, this operation is exploited in modern non-local denoising techniques [1] and in UAV navigation and imaging sensors [2]. However, there are quite many factors that make similar patch search a complicated task. Among them, one of the most important factors is noise present in all RS images. If noise is not intensive, many methods of similar patch search perform well enough. However, they start to fail if noise intensity increases and/or noise is not additive white Gaussian.