A method of hyperspectral image classification based on posterior probability SVM and MRF

In order to improve the accuracy of remote sensing image classification, this paper firstly improves the kernel function of support vector machines (SVMs), after which the Markov Random Field (MRF) stochastic model is combined with the SVM model to classify the images. The remote sensing experimental area in northwest Indiana is shot in June 1992. A VIRIS hyperspectral remote sensing images are used as an example to validate the classification algorithm, and the comparison and analysis are carried out with the traditional SVM and MRF. Experiments show that the new classification method can achieve higher classification accuracy.