A novel remote sensing image classification algorithm based on PCA and hidden Markov random field theory

Abstract

In this paper, we propose the novel remote sensing image classification algorithm based on the PCA and hidden Markov random field theory. Remote sensing image classification method based on the pattern recognition theory at home and abroad, many experts have done a lot of research work, machine learning has been in study of remote sensing image classification and information extraction has been widely used. Because of the image data is two-dimensional, so the probability of system introduced a special state, in the state relies on the observation system in horizontal and vertical two directions of the adjacent state. Therefore, we apply the expanded hidden Markov model for the image classification while the classifier task is of structure of this article from the training data estimate two-dimensional hidden Markov model. Through the experimental analysis and the corresponding test, we verify the feasibility of the algorithm.