Dimensionality reduction based on parallel factor analysis model and independent component analysis method
In hyperspectral image (HSI) analysis, dimensionality reduction is a preprocessing step for HSI classification. Independent component analysis (ICA) reduces the spectral dimension and does not utilize the spatial information of the HSI. To solve it, tensor decompositions have been successfully applied to joint noise reduction in spatial and spectral dimensions of HSIs, such as parallel factor analysis (PARAFAC). However, the PARAFAC method does not reduce the dimension in the spectral dimension. We proposed a method to improve it, which combines ICA and PARAFAC to reduce both the dimension in the spectral dimension and the noise in the spatial and spectral dimensions. The experimental results indicate that this method improves the classification compared with the previous methods.