Classification of Hyperspectral Images using Mixture of Probabilistic PCA Models

We propose a supervised classification and dimensionality reduction method for hyperspectral images. The proposed method contains a mixture of probabilistic principal component analysis (PPCA) models. Using the PPCA in the mixture model inherently provides a dimensionality reduction. Defining the mixture model to be spatially varying, we are also able to include spatial information into the classification process. In this way, the proposed mixture model allows dimensionality reduction and spectral-spatial classification of hyperspectral image at the same time. The experimental results obtained on real hyperspectral data show that the proposed method yields better classification performance compared to state of the art methods.