Hyperspectral Image Classification With  Markov Fields

This paper presents a new approach for hyperspectral imageclassification exploiting spectral-spatial information. Under the maximum a posteriori framework, we propose a supervised classification model which includes a spectral data fidelity term and a spatially adaptive Markov random field (MRF) prior in the hidden field. The data fidelity term adopted in this paper is learned from the sparse multinomial logistic regression (SMLR) classifier, while the spatially adaptive MRF prior is modeled by a spatially adaptive total variation (SpATV) regularization to enforce a spatially smooth classifier.

To further improve the classification accuracy, the true labels of training samples are fixed as an additional constraint in the proposed model. Thus, our model takes full advantage of exploiting the spatial and contextual information present in the hyperspectral image. An efficient hyperspectral image classification algorithm, named SMLR-SpATV, is then developed to solve the final proposed model using the alternating direction method of multipliers. Experimental results on real hyperspectral data sets demonstrate that the proposed approach outperforms many state-of-the-art methods in terms of the overall accuracy, average accuracy, and kappa (k) statistic.