Improving the Analysis of Hyperspectral Images Using Tensor Decomposition

With the development of remote sensing hyperspectral image analysis techniques become more important. Known methods and algorithms for pattern recognition and clustering together with others built especially for this type of data are trying to be used for dimensional reduction, spectral unmixing, decomposition, segmentation. In this paper, we approached Parafac decomposition due to the lower computation time for obtaining a model that explains very well the real data. Image segmentation techniques are applied to the abundances map to distinguish materials in an area of interest. These techniques are used to obtain a representation that will facilitate the interpretation of the information contained in the hyperspectral image.