Morphological profiles are a common approach for extracting spatial information from remote sensing hyperspectral images by extracting structural features. Other profiles can be built based on different approaches such as, for example, differential morphological profiles, or attribute profiles. Another technique used for characterizing spatial information on the images at different scales is based on computing profiles relying on edge-preserving filters such as anisotropic diffusion filters. Their main advantage is that they preserve the distinctive morphological features of the images at the cost of an iterative calculation. In this paper, the high computational cost associated to the construction of Anisotropic Diffusion Profiles (ADPs) is highly reduced. In particular, we propose a low cost computational approach for computing ADPs on Nvidia GPUs as well as a detailed characterization of the method, comparing it in terms of accuracy and structural similarity to other existing alternatives.
Keywords: Anisotropic diffusion profile, CUDA, hyperspectral, nonlinear diffusion, remote sensing